ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/libev/ev.html
Revision: 1.74
Committed: Sun Dec 9 19:42:57 2007 UTC (16 years, 5 months ago) by root
Content type: text/html
Branch: MAIN
Changes since 1.73: +20 -10 lines
Log Message:
*** empty log message ***

File Contents

# User Rev Content
1 root 1.1 <?xml version="1.0" encoding="UTF-8"?>
2     <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
3     <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en">
4     <head>
5     <title>libev</title>
6     <meta name="description" content="Pod documentation for libev" />
7     <meta name="inputfile" content="&lt;standard input&gt;" />
8     <meta name="outputfile" content="&lt;standard output&gt;" />
9 root 1.74 <meta name="created" content="Sun Dec 9 20:30:11 2007" />
10 root 1.1 <meta name="generator" content="Pod::Xhtml 1.57" />
11     <link rel="stylesheet" href="http://res.tst.eu/pod.css"/></head>
12     <body>
13     <div class="pod">
14     <!-- INDEX START -->
15     <h3 id="TOP">Index</h3>
16    
17     <ul><li><a href="#NAME">NAME</a></li>
18     <li><a href="#SYNOPSIS">SYNOPSIS</a></li>
19 root 1.54 <li><a href="#EXAMPLE_PROGRAM">EXAMPLE PROGRAM</a></li>
20 root 1.1 <li><a href="#DESCRIPTION">DESCRIPTION</a></li>
21     <li><a href="#FEATURES">FEATURES</a></li>
22     <li><a href="#CONVENTIONS">CONVENTIONS</a></li>
23 root 1.18 <li><a href="#TIME_REPRESENTATION">TIME REPRESENTATION</a></li>
24     <li><a href="#GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</a></li>
25 root 1.1 <li><a href="#FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</a></li>
26     <li><a href="#ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</a>
27 root 1.43 <ul><li><a href="#GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</a></li>
28 root 1.37 <li><a href="#ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</a></li>
29 root 1.1 </ul>
30     </li>
31     <li><a href="#WATCHER_TYPES">WATCHER TYPES</a>
32 root 1.43 <ul><li><a href="#code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</a></li>
33     <li><a href="#code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</a></li>
34     <li><a href="#code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</a></li>
35     <li><a href="#code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</a></li>
36     <li><a href="#code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</a></li>
37 root 1.48 <li><a href="#code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</a></li>
38 root 1.43 <li><a href="#code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</a></li>
39     <li><a href="#code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</a></li>
40     <li><a href="#code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</a></li>
41 root 1.50 <li><a href="#code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</a></li>
42 root 1.1 </ul>
43     </li>
44     <li><a href="#OTHER_FUNCTIONS">OTHER FUNCTIONS</a></li>
45 root 1.23 <li><a href="#LIBEVENT_EMULATION">LIBEVENT EMULATION</a></li>
46     <li><a href="#C_SUPPORT">C++ SUPPORT</a></li>
47 root 1.50 <li><a href="#MACRO_MAGIC">MACRO MAGIC</a></li>
48 root 1.40 <li><a href="#EMBEDDING">EMBEDDING</a>
49     <ul><li><a href="#FILESETS">FILESETS</a>
50     <ul><li><a href="#CORE_EVENT_LOOP">CORE EVENT LOOP</a></li>
51     <li><a href="#LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</a></li>
52     <li><a href="#AUTOCONF_SUPPORT">AUTOCONF SUPPORT</a></li>
53     </ul>
54     </li>
55     <li><a href="#PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</a></li>
56     <li><a href="#EXAMPLES">EXAMPLES</a></li>
57     </ul>
58     </li>
59 root 1.47 <li><a href="#COMPLEXITIES">COMPLEXITIES</a></li>
60 root 1.1 <li><a href="#AUTHOR">AUTHOR</a>
61     </li>
62     </ul><hr />
63     <!-- INDEX END -->
64    
65 root 1.55 <h1 id="NAME">NAME</h1>
66 root 1.1 <div id="NAME_CONTENT">
67     <p>libev - a high performance full-featured event loop written in C</p>
68    
69     </div>
70 root 1.55 <h1 id="SYNOPSIS">SYNOPSIS</h1>
71 root 1.1 <div id="SYNOPSIS_CONTENT">
72 root 1.54 <pre> #include &lt;ev.h&gt;
73    
74     </pre>
75    
76     </div>
77 root 1.55 <h1 id="EXAMPLE_PROGRAM">EXAMPLE PROGRAM</h1>
78 root 1.54 <div id="EXAMPLE_PROGRAM_CONTENT">
79     <pre> #include &lt;ev.h&gt;
80 root 1.53
81     ev_io stdin_watcher;
82     ev_timer timeout_watcher;
83    
84     /* called when data readable on stdin */
85     static void
86     stdin_cb (EV_P_ struct ev_io *w, int revents)
87     {
88     /* puts (&quot;stdin ready&quot;); */
89     ev_io_stop (EV_A_ w); /* just a syntax example */
90     ev_unloop (EV_A_ EVUNLOOP_ALL); /* leave all loop calls */
91     }
92    
93     static void
94     timeout_cb (EV_P_ struct ev_timer *w, int revents)
95     {
96     /* puts (&quot;timeout&quot;); */
97     ev_unloop (EV_A_ EVUNLOOP_ONE); /* leave one loop call */
98     }
99    
100     int
101     main (void)
102     {
103     struct ev_loop *loop = ev_default_loop (0);
104    
105     /* initialise an io watcher, then start it */
106     ev_io_init (&amp;stdin_watcher, stdin_cb, /*STDIN_FILENO*/ 0, EV_READ);
107     ev_io_start (loop, &amp;stdin_watcher);
108    
109     /* simple non-repeating 5.5 second timeout */
110     ev_timer_init (&amp;timeout_watcher, timeout_cb, 5.5, 0.);
111     ev_timer_start (loop, &amp;timeout_watcher);
112    
113     /* loop till timeout or data ready */
114     ev_loop (loop, 0);
115    
116     return 0;
117     }
118 root 1.1
119     </pre>
120    
121     </div>
122 root 1.55 <h1 id="DESCRIPTION">DESCRIPTION</h1>
123 root 1.1 <div id="DESCRIPTION_CONTENT">
124 root 1.66 <p>The newest version of this document is also available as a html-formatted
125     web page you might find easier to navigate when reading it for the first
126     time: <a href="http://cvs.schmorp.de/libev/ev.html">http://cvs.schmorp.de/libev/ev.html</a>.</p>
127 root 1.1 <p>Libev is an event loop: you register interest in certain events (such as a
128     file descriptor being readable or a timeout occuring), and it will manage
129 root 1.4 these event sources and provide your program with events.</p>
130 root 1.1 <p>To do this, it must take more or less complete control over your process
131     (or thread) by executing the <i>event loop</i> handler, and will then
132     communicate events via a callback mechanism.</p>
133     <p>You register interest in certain events by registering so-called <i>event
134     watchers</i>, which are relatively small C structures you initialise with the
135     details of the event, and then hand it over to libev by <i>starting</i> the
136     watcher.</p>
137    
138     </div>
139 root 1.55 <h1 id="FEATURES">FEATURES</h1>
140 root 1.1 <div id="FEATURES_CONTENT">
141 root 1.58 <p>Libev supports <code>select</code>, <code>poll</code>, the Linux-specific <code>epoll</code>, the
142     BSD-specific <code>kqueue</code> and the Solaris-specific event port mechanisms
143     for file descriptor events (<code>ev_io</code>), the Linux <code>inotify</code> interface
144     (for <code>ev_stat</code>), relative timers (<code>ev_timer</code>), absolute timers
145     with customised rescheduling (<code>ev_periodic</code>), synchronous signals
146     (<code>ev_signal</code>), process status change events (<code>ev_child</code>), and event
147     watchers dealing with the event loop mechanism itself (<code>ev_idle</code>,
148 root 1.54 <code>ev_embed</code>, <code>ev_prepare</code> and <code>ev_check</code> watchers) as well as
149     file watchers (<code>ev_stat</code>) and even limited support for fork events
150     (<code>ev_fork</code>).</p>
151     <p>It also is quite fast (see this
152     <a href="http://libev.schmorp.de/bench.html">benchmark</a> comparing it to libevent
153     for example).</p>
154 root 1.1
155     </div>
156 root 1.55 <h1 id="CONVENTIONS">CONVENTIONS</h1>
157 root 1.1 <div id="CONVENTIONS_CONTENT">
158 root 1.54 <p>Libev is very configurable. In this manual the default configuration will
159     be described, which supports multiple event loops. For more info about
160     various configuration options please have a look at <strong>EMBED</strong> section in
161     this manual. If libev was configured without support for multiple event
162     loops, then all functions taking an initial argument of name <code>loop</code>
163     (which is always of type <code>struct ev_loop *</code>) will not have this argument.</p>
164 root 1.1
165     </div>
166 root 1.55 <h1 id="TIME_REPRESENTATION">TIME REPRESENTATION</h1>
167 root 1.18 <div id="TIME_REPRESENTATION_CONTENT">
168 root 1.2 <p>Libev represents time as a single floating point number, representing the
169     (fractional) number of seconds since the (POSIX) epoch (somewhere near
170     the beginning of 1970, details are complicated, don't ask). This type is
171 root 1.1 called <code>ev_tstamp</code>, which is what you should use too. It usually aliases
172 root 1.35 to the <code>double</code> type in C, and when you need to do any calculations on
173     it, you should treat it as such.</p>
174    
175 root 1.18 </div>
176 root 1.55 <h1 id="GLOBAL_FUNCTIONS">GLOBAL FUNCTIONS</h1>
177 root 1.18 <div id="GLOBAL_FUNCTIONS_CONTENT">
178 root 1.21 <p>These functions can be called anytime, even before initialising the
179     library in any way.</p>
180 root 1.1 <dl>
181     <dt>ev_tstamp ev_time ()</dt>
182     <dd>
183 root 1.28 <p>Returns the current time as libev would use it. Please note that the
184     <code>ev_now</code> function is usually faster and also often returns the timestamp
185     you actually want to know.</p>
186 root 1.1 </dd>
187     <dt>int ev_version_major ()</dt>
188     <dt>int ev_version_minor ()</dt>
189     <dd>
190     <p>You can find out the major and minor version numbers of the library
191     you linked against by calling the functions <code>ev_version_major</code> and
192     <code>ev_version_minor</code>. If you want, you can compare against the global
193     symbols <code>EV_VERSION_MAJOR</code> and <code>EV_VERSION_MINOR</code>, which specify the
194     version of the library your program was compiled against.</p>
195 root 1.9 <p>Usually, it's a good idea to terminate if the major versions mismatch,
196 root 1.1 as this indicates an incompatible change. Minor versions are usually
197     compatible to older versions, so a larger minor version alone is usually
198     not a problem.</p>
199 root 1.54 <p>Example: Make sure we haven't accidentally been linked against the wrong
200     version.</p>
201 root 1.35 <pre> assert ((&quot;libev version mismatch&quot;,
202     ev_version_major () == EV_VERSION_MAJOR
203     &amp;&amp; ev_version_minor () &gt;= EV_VERSION_MINOR));
204    
205     </pre>
206 root 1.1 </dd>
207 root 1.32 <dt>unsigned int ev_supported_backends ()</dt>
208     <dd>
209     <p>Return the set of all backends (i.e. their corresponding <code>EV_BACKEND_*</code>
210     value) compiled into this binary of libev (independent of their
211     availability on the system you are running on). See <code>ev_default_loop</code> for
212     a description of the set values.</p>
213 root 1.35 <p>Example: make sure we have the epoll method, because yeah this is cool and
214     a must have and can we have a torrent of it please!!!11</p>
215     <pre> assert ((&quot;sorry, no epoll, no sex&quot;,
216     ev_supported_backends () &amp; EVBACKEND_EPOLL));
217    
218     </pre>
219 root 1.32 </dd>
220     <dt>unsigned int ev_recommended_backends ()</dt>
221     <dd>
222     <p>Return the set of all backends compiled into this binary of libev and also
223     recommended for this platform. This set is often smaller than the one
224     returned by <code>ev_supported_backends</code>, as for example kqueue is broken on
225     most BSDs and will not be autodetected unless you explicitly request it
226     (assuming you know what you are doing). This is the set of backends that
227 root 1.34 libev will probe for if you specify no backends explicitly.</p>
228 root 1.32 </dd>
229 root 1.36 <dt>unsigned int ev_embeddable_backends ()</dt>
230     <dd>
231     <p>Returns the set of backends that are embeddable in other event loops. This
232     is the theoretical, all-platform, value. To find which backends
233     might be supported on the current system, you would need to look at
234     <code>ev_embeddable_backends () &amp; ev_supported_backends ()</code>, likewise for
235     recommended ones.</p>
236     <p>See the description of <code>ev_embed</code> watchers for more info.</p>
237     </dd>
238 root 1.59 <dt>ev_set_allocator (void *(*cb)(void *ptr, long size))</dt>
239 root 1.1 <dd>
240 root 1.59 <p>Sets the allocation function to use (the prototype is similar - the
241     semantics is identical - to the realloc C function). It is used to
242     allocate and free memory (no surprises here). If it returns zero when
243     memory needs to be allocated, the library might abort or take some
244     potentially destructive action. The default is your system realloc
245     function.</p>
246 root 1.1 <p>You could override this function in high-availability programs to, say,
247     free some memory if it cannot allocate memory, to use a special allocator,
248     or even to sleep a while and retry until some memory is available.</p>
249 root 1.54 <p>Example: Replace the libev allocator with one that waits a bit and then
250     retries).</p>
251 root 1.35 <pre> static void *
252 root 1.52 persistent_realloc (void *ptr, size_t size)
253 root 1.35 {
254     for (;;)
255     {
256     void *newptr = realloc (ptr, size);
257    
258     if (newptr)
259     return newptr;
260    
261     sleep (60);
262     }
263     }
264    
265     ...
266     ev_set_allocator (persistent_realloc);
267    
268     </pre>
269 root 1.1 </dd>
270     <dt>ev_set_syserr_cb (void (*cb)(const char *msg));</dt>
271     <dd>
272     <p>Set the callback function to call on a retryable syscall error (such
273     as failed select, poll, epoll_wait). The message is a printable string
274     indicating the system call or subsystem causing the problem. If this
275     callback is set, then libev will expect it to remedy the sitution, no
276 root 1.7 matter what, when it returns. That is, libev will generally retry the
277 root 1.1 requested operation, or, if the condition doesn't go away, do bad stuff
278     (such as abort).</p>
279 root 1.54 <p>Example: This is basically the same thing that libev does internally, too.</p>
280 root 1.35 <pre> static void
281     fatal_error (const char *msg)
282     {
283     perror (msg);
284     abort ();
285     }
286    
287     ...
288     ev_set_syserr_cb (fatal_error);
289    
290     </pre>
291 root 1.1 </dd>
292     </dl>
293    
294     </div>
295 root 1.55 <h1 id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP">FUNCTIONS CONTROLLING THE EVENT LOOP</h1>
296 root 1.1 <div id="FUNCTIONS_CONTROLLING_THE_EVENT_LOOP-2">
297     <p>An event loop is described by a <code>struct ev_loop *</code>. The library knows two
298     types of such loops, the <i>default</i> loop, which supports signals and child
299     events, and dynamically created loops which do not.</p>
300     <p>If you use threads, a common model is to run the default event loop
301 root 1.18 in your main thread (or in a separate thread) and for each thread you
302 root 1.7 create, you also create another event loop. Libev itself does no locking
303     whatsoever, so if you mix calls to the same event loop in different
304     threads, make sure you lock (this is usually a bad idea, though, even if
305 root 1.9 done correctly, because it's hideous and inefficient).</p>
306 root 1.1 <dl>
307     <dt>struct ev_loop *ev_default_loop (unsigned int flags)</dt>
308     <dd>
309     <p>This will initialise the default event loop if it hasn't been initialised
310     yet and return it. If the default loop could not be initialised, returns
311     false. If it already was initialised it simply returns it (and ignores the
312 root 1.32 flags. If that is troubling you, check <code>ev_backend ()</code> afterwards).</p>
313 root 1.1 <p>If you don't know what event loop to use, use the one returned from this
314     function.</p>
315     <p>The flags argument can be used to specify special behaviour or specific
316 root 1.34 backends to use, and is usually specified as <code>0</code> (or <code>EVFLAG_AUTO</code>).</p>
317     <p>The following flags are supported:</p>
318 root 1.1 <p>
319     <dl>
320 root 1.10 <dt><code>EVFLAG_AUTO</code></dt>
321 root 1.1 <dd>
322 root 1.9 <p>The default flags value. Use this if you have no clue (it's the right
323 root 1.1 thing, believe me).</p>
324     </dd>
325 root 1.10 <dt><code>EVFLAG_NOENV</code></dt>
326 root 1.1 <dd>
327 root 1.8 <p>If this flag bit is ored into the flag value (or the program runs setuid
328     or setgid) then libev will <i>not</i> look at the environment variable
329     <code>LIBEV_FLAGS</code>. Otherwise (the default), this environment variable will
330     override the flags completely if it is found in the environment. This is
331     useful to try out specific backends to test their performance, or to work
332     around bugs.</p>
333 root 1.1 </dd>
334 root 1.62 <dt><code>EVFLAG_FORKCHECK</code></dt>
335     <dd>
336     <p>Instead of calling <code>ev_default_fork</code> or <code>ev_loop_fork</code> manually after
337     a fork, you can also make libev check for a fork in each iteration by
338     enabling this flag.</p>
339     <p>This works by calling <code>getpid ()</code> on every iteration of the loop,
340     and thus this might slow down your event loop if you do a lot of loop
341 root 1.64 iterations and little real work, but is usually not noticeable (on my
342 root 1.62 Linux system for example, <code>getpid</code> is actually a simple 5-insn sequence
343     without a syscall and thus <i>very</i> fast, but my Linux system also has
344     <code>pthread_atfork</code> which is even faster).</p>
345     <p>The big advantage of this flag is that you can forget about fork (and
346     forget about forgetting to tell libev about forking) when you use this
347     flag.</p>
348     <p>This flag setting cannot be overriden or specified in the <code>LIBEV_FLAGS</code>
349     environment variable.</p>
350     </dd>
351 root 1.32 <dt><code>EVBACKEND_SELECT</code> (value 1, portable select backend)</dt>
352 root 1.29 <dd>
353     <p>This is your standard select(2) backend. Not <i>completely</i> standard, as
354     libev tries to roll its own fd_set with no limits on the number of fds,
355     but if that fails, expect a fairly low limit on the number of fds when
356     using this backend. It doesn't scale too well (O(highest_fd)), but its usually
357     the fastest backend for a low number of fds.</p>
358     </dd>
359 root 1.32 <dt><code>EVBACKEND_POLL</code> (value 2, poll backend, available everywhere except on windows)</dt>
360 root 1.29 <dd>
361     <p>And this is your standard poll(2) backend. It's more complicated than
362     select, but handles sparse fds better and has no artificial limit on the
363     number of fds you can use (except it will slow down considerably with a
364     lot of inactive fds). It scales similarly to select, i.e. O(total_fds).</p>
365     </dd>
366 root 1.32 <dt><code>EVBACKEND_EPOLL</code> (value 4, Linux)</dt>
367 root 1.29 <dd>
368     <p>For few fds, this backend is a bit little slower than poll and select,
369     but it scales phenomenally better. While poll and select usually scale like
370     O(total_fds) where n is the total number of fds (or the highest fd), epoll scales
371     either O(1) or O(active_fds).</p>
372     <p>While stopping and starting an I/O watcher in the same iteration will
373     result in some caching, there is still a syscall per such incident
374     (because the fd could point to a different file description now), so its
375     best to avoid that. Also, dup()ed file descriptors might not work very
376     well if you register events for both fds.</p>
377 root 1.33 <p>Please note that epoll sometimes generates spurious notifications, so you
378     need to use non-blocking I/O or other means to avoid blocking when no data
379     (or space) is available.</p>
380 root 1.29 </dd>
381 root 1.32 <dt><code>EVBACKEND_KQUEUE</code> (value 8, most BSD clones)</dt>
382 root 1.29 <dd>
383     <p>Kqueue deserves special mention, as at the time of this writing, it
384     was broken on all BSDs except NetBSD (usually it doesn't work with
385     anything but sockets and pipes, except on Darwin, where of course its
386 root 1.34 completely useless). For this reason its not being &quot;autodetected&quot;
387     unless you explicitly specify it explicitly in the flags (i.e. using
388     <code>EVBACKEND_KQUEUE</code>).</p>
389 root 1.29 <p>It scales in the same way as the epoll backend, but the interface to the
390     kernel is more efficient (which says nothing about its actual speed, of
391     course). While starting and stopping an I/O watcher does not cause an
392     extra syscall as with epoll, it still adds up to four event changes per
393     incident, so its best to avoid that.</p>
394     </dd>
395 root 1.32 <dt><code>EVBACKEND_DEVPOLL</code> (value 16, Solaris 8)</dt>
396 root 1.29 <dd>
397     <p>This is not implemented yet (and might never be).</p>
398     </dd>
399 root 1.32 <dt><code>EVBACKEND_PORT</code> (value 32, Solaris 10)</dt>
400 root 1.29 <dd>
401     <p>This uses the Solaris 10 port mechanism. As with everything on Solaris,
402     it's really slow, but it still scales very well (O(active_fds)).</p>
403 root 1.33 <p>Please note that solaris ports can result in a lot of spurious
404     notifications, so you need to use non-blocking I/O or other means to avoid
405     blocking when no data (or space) is available.</p>
406 root 1.29 </dd>
407 root 1.32 <dt><code>EVBACKEND_ALL</code></dt>
408 root 1.29 <dd>
409 root 1.30 <p>Try all backends (even potentially broken ones that wouldn't be tried
410     with <code>EVFLAG_AUTO</code>). Since this is a mask, you can do stuff such as
411 root 1.32 <code>EVBACKEND_ALL &amp; ~EVBACKEND_KQUEUE</code>.</p>
412 root 1.1 </dd>
413     </dl>
414     </p>
415 root 1.29 <p>If one or more of these are ored into the flags value, then only these
416     backends will be tried (in the reverse order as given here). If none are
417     specified, most compiled-in backend will be tried, usually in reverse
418     order of their flag values :)</p>
419 root 1.34 <p>The most typical usage is like this:</p>
420     <pre> if (!ev_default_loop (0))
421     fatal (&quot;could not initialise libev, bad $LIBEV_FLAGS in environment?&quot;);
422    
423     </pre>
424     <p>Restrict libev to the select and poll backends, and do not allow
425     environment settings to be taken into account:</p>
426     <pre> ev_default_loop (EVBACKEND_POLL | EVBACKEND_SELECT | EVFLAG_NOENV);
427    
428     </pre>
429     <p>Use whatever libev has to offer, but make sure that kqueue is used if
430     available (warning, breaks stuff, best use only with your own private
431     event loop and only if you know the OS supports your types of fds):</p>
432     <pre> ev_default_loop (ev_recommended_backends () | EVBACKEND_KQUEUE);
433    
434     </pre>
435 root 1.1 </dd>
436     <dt>struct ev_loop *ev_loop_new (unsigned int flags)</dt>
437     <dd>
438     <p>Similar to <code>ev_default_loop</code>, but always creates a new event loop that is
439     always distinct from the default loop. Unlike the default loop, it cannot
440     handle signal and child watchers, and attempts to do so will be greeted by
441     undefined behaviour (or a failed assertion if assertions are enabled).</p>
442 root 1.54 <p>Example: Try to create a event loop that uses epoll and nothing else.</p>
443 root 1.35 <pre> struct ev_loop *epoller = ev_loop_new (EVBACKEND_EPOLL | EVFLAG_NOENV);
444     if (!epoller)
445     fatal (&quot;no epoll found here, maybe it hides under your chair&quot;);
446    
447     </pre>
448 root 1.1 </dd>
449     <dt>ev_default_destroy ()</dt>
450     <dd>
451     <p>Destroys the default loop again (frees all memory and kernel state
452 root 1.38 etc.). None of the active event watchers will be stopped in the normal
453     sense, so e.g. <code>ev_is_active</code> might still return true. It is your
454     responsibility to either stop all watchers cleanly yoursef <i>before</i>
455     calling this function, or cope with the fact afterwards (which is usually
456     the easiest thing, youc na just ignore the watchers and/or <code>free ()</code> them
457     for example).</p>
458 root 1.1 </dd>
459     <dt>ev_loop_destroy (loop)</dt>
460     <dd>
461     <p>Like <code>ev_default_destroy</code>, but destroys an event loop created by an
462     earlier call to <code>ev_loop_new</code>.</p>
463     </dd>
464     <dt>ev_default_fork ()</dt>
465     <dd>
466     <p>This function reinitialises the kernel state for backends that have
467     one. Despite the name, you can call it anytime, but it makes most sense
468     after forking, in either the parent or child process (or both, but that
469     again makes little sense).</p>
470 root 1.31 <p>You <i>must</i> call this function in the child process after forking if and
471     only if you want to use the event library in both processes. If you just
472     fork+exec, you don't have to call it.</p>
473 root 1.9 <p>The function itself is quite fast and it's usually not a problem to call
474 root 1.1 it just in case after a fork. To make this easy, the function will fit in
475     quite nicely into a call to <code>pthread_atfork</code>:</p>
476     <pre> pthread_atfork (0, 0, ev_default_fork);
477    
478     </pre>
479 root 1.32 <p>At the moment, <code>EVBACKEND_SELECT</code> and <code>EVBACKEND_POLL</code> are safe to use
480     without calling this function, so if you force one of those backends you
481     do not need to care.</p>
482 root 1.1 </dd>
483     <dt>ev_loop_fork (loop)</dt>
484     <dd>
485     <p>Like <code>ev_default_fork</code>, but acts on an event loop created by
486     <code>ev_loop_new</code>. Yes, you have to call this on every allocated event loop
487     after fork, and how you do this is entirely your own problem.</p>
488     </dd>
489 root 1.64 <dt>unsigned int ev_loop_count (loop)</dt>
490     <dd>
491     <p>Returns the count of loop iterations for the loop, which is identical to
492     the number of times libev did poll for new events. It starts at <code>0</code> and
493     happily wraps around with enough iterations.</p>
494     <p>This value can sometimes be useful as a generation counter of sorts (it
495     &quot;ticks&quot; the number of loop iterations), as it roughly corresponds with
496     <code>ev_prepare</code> and <code>ev_check</code> calls.</p>
497     </dd>
498 root 1.32 <dt>unsigned int ev_backend (loop)</dt>
499 root 1.1 <dd>
500 root 1.32 <p>Returns one of the <code>EVBACKEND_*</code> flags indicating the event backend in
501 root 1.1 use.</p>
502     </dd>
503 root 1.9 <dt>ev_tstamp ev_now (loop)</dt>
504 root 1.1 <dd>
505     <p>Returns the current &quot;event loop time&quot;, which is the time the event loop
506 root 1.35 received events and started processing them. This timestamp does not
507     change as long as callbacks are being processed, and this is also the base
508     time used for relative timers. You can treat it as the timestamp of the
509     event occuring (or more correctly, libev finding out about it).</p>
510 root 1.1 </dd>
511     <dt>ev_loop (loop, int flags)</dt>
512     <dd>
513     <p>Finally, this is it, the event handler. This function usually is called
514     after you initialised all your watchers and you want to start handling
515     events.</p>
516 root 1.34 <p>If the flags argument is specified as <code>0</code>, it will not return until
517     either no event watchers are active anymore or <code>ev_unloop</code> was called.</p>
518 root 1.35 <p>Please note that an explicit <code>ev_unloop</code> is usually better than
519     relying on all watchers to be stopped when deciding when a program has
520     finished (especially in interactive programs), but having a program that
521     automatically loops as long as it has to and no longer by virtue of
522     relying on its watchers stopping correctly is a thing of beauty.</p>
523 root 1.1 <p>A flags value of <code>EVLOOP_NONBLOCK</code> will look for new events, will handle
524     those events and any outstanding ones, but will not block your process in
525 root 1.9 case there are no events and will return after one iteration of the loop.</p>
526 root 1.1 <p>A flags value of <code>EVLOOP_ONESHOT</code> will look for new events (waiting if
527     neccessary) and will handle those and any outstanding ones. It will block
528 root 1.9 your process until at least one new event arrives, and will return after
529 root 1.34 one iteration of the loop. This is useful if you are waiting for some
530     external event in conjunction with something not expressible using other
531     libev watchers. However, a pair of <code>ev_prepare</code>/<code>ev_check</code> watchers is
532     usually a better approach for this kind of thing.</p>
533     <p>Here are the gory details of what <code>ev_loop</code> does:</p>
534 root 1.73 <pre> - Before the first iteration, call any pending watchers.
535     * If there are no active watchers (reference count is zero), return.
536     - Queue all prepare watchers and then call all outstanding watchers.
537 root 1.34 - If we have been forked, recreate the kernel state.
538     - Update the kernel state with all outstanding changes.
539     - Update the &quot;event loop time&quot;.
540     - Calculate for how long to block.
541     - Block the process, waiting for any events.
542     - Queue all outstanding I/O (fd) events.
543     - Update the &quot;event loop time&quot; and do time jump handling.
544     - Queue all outstanding timers.
545     - Queue all outstanding periodics.
546     - If no events are pending now, queue all idle watchers.
547     - Queue all check watchers.
548     - Call all queued watchers in reverse order (i.e. check watchers first).
549     Signals and child watchers are implemented as I/O watchers, and will
550     be handled here by queueing them when their watcher gets executed.
551     - If ev_unloop has been called or EVLOOP_ONESHOT or EVLOOP_NONBLOCK
552     were used, return, otherwise continue with step *.
553 root 1.28
554     </pre>
555 root 1.54 <p>Example: Queue some jobs and then loop until no events are outsanding
556 root 1.35 anymore.</p>
557     <pre> ... queue jobs here, make sure they register event watchers as long
558     ... as they still have work to do (even an idle watcher will do..)
559     ev_loop (my_loop, 0);
560     ... jobs done. yeah!
561    
562     </pre>
563 root 1.1 </dd>
564     <dt>ev_unloop (loop, how)</dt>
565     <dd>
566 root 1.9 <p>Can be used to make a call to <code>ev_loop</code> return early (but only after it
567     has processed all outstanding events). The <code>how</code> argument must be either
568 root 1.27 <code>EVUNLOOP_ONE</code>, which will make the innermost <code>ev_loop</code> call return, or
569 root 1.9 <code>EVUNLOOP_ALL</code>, which will make all nested <code>ev_loop</code> calls return.</p>
570 root 1.1 </dd>
571     <dt>ev_ref (loop)</dt>
572     <dt>ev_unref (loop)</dt>
573     <dd>
574 root 1.9 <p>Ref/unref can be used to add or remove a reference count on the event
575     loop: Every watcher keeps one reference, and as long as the reference
576     count is nonzero, <code>ev_loop</code> will not return on its own. If you have
577     a watcher you never unregister that should not keep <code>ev_loop</code> from
578     returning, ev_unref() after starting, and ev_ref() before stopping it. For
579     example, libev itself uses this for its internal signal pipe: It is not
580     visible to the libev user and should not keep <code>ev_loop</code> from exiting if
581     no event watchers registered by it are active. It is also an excellent
582     way to do this for generic recurring timers or from within third-party
583     libraries. Just remember to <i>unref after start</i> and <i>ref before stop</i>.</p>
584 root 1.54 <p>Example: Create a signal watcher, but keep it from keeping <code>ev_loop</code>
585 root 1.35 running when nothing else is active.</p>
586 root 1.54 <pre> struct ev_signal exitsig;
587 root 1.35 ev_signal_init (&amp;exitsig, sig_cb, SIGINT);
588 root 1.54 ev_signal_start (loop, &amp;exitsig);
589     evf_unref (loop);
590 root 1.35
591     </pre>
592 root 1.54 <p>Example: For some weird reason, unregister the above signal handler again.</p>
593     <pre> ev_ref (loop);
594     ev_signal_stop (loop, &amp;exitsig);
595 root 1.35
596     </pre>
597 root 1.1 </dd>
598     </dl>
599    
600 root 1.43
601    
602    
603    
604 root 1.1 </div>
605 root 1.55 <h1 id="ANATOMY_OF_A_WATCHER">ANATOMY OF A WATCHER</h1>
606 root 1.1 <div id="ANATOMY_OF_A_WATCHER_CONTENT">
607     <p>A watcher is a structure that you create and register to record your
608     interest in some event. For instance, if you want to wait for STDIN to
609 root 1.10 become readable, you would create an <code>ev_io</code> watcher for that:</p>
610 root 1.1 <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w, int revents)
611     {
612     ev_io_stop (w);
613     ev_unloop (loop, EVUNLOOP_ALL);
614     }
615    
616     struct ev_loop *loop = ev_default_loop (0);
617     struct ev_io stdin_watcher;
618     ev_init (&amp;stdin_watcher, my_cb);
619     ev_io_set (&amp;stdin_watcher, STDIN_FILENO, EV_READ);
620     ev_io_start (loop, &amp;stdin_watcher);
621     ev_loop (loop, 0);
622    
623     </pre>
624     <p>As you can see, you are responsible for allocating the memory for your
625     watcher structures (and it is usually a bad idea to do this on the stack,
626     although this can sometimes be quite valid).</p>
627     <p>Each watcher structure must be initialised by a call to <code>ev_init
628     (watcher *, callback)</code>, which expects a callback to be provided. This
629     callback gets invoked each time the event occurs (or, in the case of io
630     watchers, each time the event loop detects that the file descriptor given
631     is readable and/or writable).</p>
632     <p>Each watcher type has its own <code>ev_&lt;type&gt;_set (watcher *, ...)</code> macro
633     with arguments specific to this watcher type. There is also a macro
634     to combine initialisation and setting in one call: <code>ev_&lt;type&gt;_init
635     (watcher *, callback, ...)</code>.</p>
636     <p>To make the watcher actually watch out for events, you have to start it
637     with a watcher-specific start function (<code>ev_&lt;type&gt;_start (loop, watcher
638     *)</code>), and you can stop watching for events at any time by calling the
639     corresponding stop function (<code>ev_&lt;type&gt;_stop (loop, watcher *)</code>.</p>
640     <p>As long as your watcher is active (has been started but not stopped) you
641     must not touch the values stored in it. Most specifically you must never
642 root 1.37 reinitialise it or call its <code>set</code> macro.</p>
643 root 1.1 <p>Each and every callback receives the event loop pointer as first, the
644     registered watcher structure as second, and a bitset of received events as
645     third argument.</p>
646 root 1.14 <p>The received events usually include a single bit per event type received
647 root 1.1 (you can receive multiple events at the same time). The possible bit masks
648     are:</p>
649     <dl>
650 root 1.10 <dt><code>EV_READ</code></dt>
651     <dt><code>EV_WRITE</code></dt>
652 root 1.1 <dd>
653 root 1.10 <p>The file descriptor in the <code>ev_io</code> watcher has become readable and/or
654 root 1.1 writable.</p>
655     </dd>
656 root 1.10 <dt><code>EV_TIMEOUT</code></dt>
657 root 1.1 <dd>
658 root 1.10 <p>The <code>ev_timer</code> watcher has timed out.</p>
659 root 1.1 </dd>
660 root 1.10 <dt><code>EV_PERIODIC</code></dt>
661 root 1.1 <dd>
662 root 1.10 <p>The <code>ev_periodic</code> watcher has timed out.</p>
663 root 1.1 </dd>
664 root 1.10 <dt><code>EV_SIGNAL</code></dt>
665 root 1.1 <dd>
666 root 1.10 <p>The signal specified in the <code>ev_signal</code> watcher has been received by a thread.</p>
667 root 1.1 </dd>
668 root 1.10 <dt><code>EV_CHILD</code></dt>
669 root 1.1 <dd>
670 root 1.10 <p>The pid specified in the <code>ev_child</code> watcher has received a status change.</p>
671 root 1.1 </dd>
672 root 1.48 <dt><code>EV_STAT</code></dt>
673     <dd>
674     <p>The path specified in the <code>ev_stat</code> watcher changed its attributes somehow.</p>
675     </dd>
676 root 1.10 <dt><code>EV_IDLE</code></dt>
677 root 1.1 <dd>
678 root 1.10 <p>The <code>ev_idle</code> watcher has determined that you have nothing better to do.</p>
679 root 1.1 </dd>
680 root 1.10 <dt><code>EV_PREPARE</code></dt>
681     <dt><code>EV_CHECK</code></dt>
682 root 1.1 <dd>
683 root 1.10 <p>All <code>ev_prepare</code> watchers are invoked just <i>before</i> <code>ev_loop</code> starts
684     to gather new events, and all <code>ev_check</code> watchers are invoked just after
685 root 1.1 <code>ev_loop</code> has gathered them, but before it invokes any callbacks for any
686     received events. Callbacks of both watcher types can start and stop as
687     many watchers as they want, and all of them will be taken into account
688 root 1.10 (for example, a <code>ev_prepare</code> watcher might start an idle watcher to keep
689 root 1.1 <code>ev_loop</code> from blocking).</p>
690     </dd>
691 root 1.50 <dt><code>EV_EMBED</code></dt>
692     <dd>
693     <p>The embedded event loop specified in the <code>ev_embed</code> watcher needs attention.</p>
694     </dd>
695     <dt><code>EV_FORK</code></dt>
696     <dd>
697     <p>The event loop has been resumed in the child process after fork (see
698     <code>ev_fork</code>).</p>
699     </dd>
700 root 1.10 <dt><code>EV_ERROR</code></dt>
701 root 1.1 <dd>
702     <p>An unspecified error has occured, the watcher has been stopped. This might
703     happen because the watcher could not be properly started because libev
704     ran out of memory, a file descriptor was found to be closed or any other
705     problem. You best act on it by reporting the problem and somehow coping
706     with the watcher being stopped.</p>
707     <p>Libev will usually signal a few &quot;dummy&quot; events together with an error,
708     for example it might indicate that a fd is readable or writable, and if
709     your callbacks is well-written it can just attempt the operation and cope
710     with the error from read() or write(). This will not work in multithreaded
711     programs, though, so beware.</p>
712     </dd>
713     </dl>
714    
715     </div>
716 root 1.43 <h2 id="GENERIC_WATCHER_FUNCTIONS">GENERIC WATCHER FUNCTIONS</h2>
717     <div id="GENERIC_WATCHER_FUNCTIONS_CONTENT">
718 root 1.37 <p>In the following description, <code>TYPE</code> stands for the watcher type,
719     e.g. <code>timer</code> for <code>ev_timer</code> watchers and <code>io</code> for <code>ev_io</code> watchers.</p>
720     <dl>
721     <dt><code>ev_init</code> (ev_TYPE *watcher, callback)</dt>
722     <dd>
723     <p>This macro initialises the generic portion of a watcher. The contents
724     of the watcher object can be arbitrary (so <code>malloc</code> will do). Only
725     the generic parts of the watcher are initialised, you <i>need</i> to call
726     the type-specific <code>ev_TYPE_set</code> macro afterwards to initialise the
727     type-specific parts. For each type there is also a <code>ev_TYPE_init</code> macro
728     which rolls both calls into one.</p>
729     <p>You can reinitialise a watcher at any time as long as it has been stopped
730     (or never started) and there are no pending events outstanding.</p>
731 root 1.43 <p>The callback is always of type <code>void (*)(ev_loop *loop, ev_TYPE *watcher,
732 root 1.37 int revents)</code>.</p>
733     </dd>
734     <dt><code>ev_TYPE_set</code> (ev_TYPE *, [args])</dt>
735     <dd>
736     <p>This macro initialises the type-specific parts of a watcher. You need to
737     call <code>ev_init</code> at least once before you call this macro, but you can
738     call <code>ev_TYPE_set</code> any number of times. You must not, however, call this
739     macro on a watcher that is active (it can be pending, however, which is a
740     difference to the <code>ev_init</code> macro).</p>
741     <p>Although some watcher types do not have type-specific arguments
742     (e.g. <code>ev_prepare</code>) you still need to call its <code>set</code> macro.</p>
743     </dd>
744     <dt><code>ev_TYPE_init</code> (ev_TYPE *watcher, callback, [args])</dt>
745     <dd>
746     <p>This convinience macro rolls both <code>ev_init</code> and <code>ev_TYPE_set</code> macro
747     calls into a single call. This is the most convinient method to initialise
748     a watcher. The same limitations apply, of course.</p>
749     </dd>
750     <dt><code>ev_TYPE_start</code> (loop *, ev_TYPE *watcher)</dt>
751     <dd>
752     <p>Starts (activates) the given watcher. Only active watchers will receive
753     events. If the watcher is already active nothing will happen.</p>
754     </dd>
755     <dt><code>ev_TYPE_stop</code> (loop *, ev_TYPE *watcher)</dt>
756     <dd>
757     <p>Stops the given watcher again (if active) and clears the pending
758     status. It is possible that stopped watchers are pending (for example,
759     non-repeating timers are being stopped when they become pending), but
760     <code>ev_TYPE_stop</code> ensures that the watcher is neither active nor pending. If
761     you want to free or reuse the memory used by the watcher it is therefore a
762     good idea to always call its <code>ev_TYPE_stop</code> function.</p>
763     </dd>
764     <dt>bool ev_is_active (ev_TYPE *watcher)</dt>
765     <dd>
766     <p>Returns a true value iff the watcher is active (i.e. it has been started
767     and not yet been stopped). As long as a watcher is active you must not modify
768     it.</p>
769     </dd>
770     <dt>bool ev_is_pending (ev_TYPE *watcher)</dt>
771     <dd>
772     <p>Returns a true value iff the watcher is pending, (i.e. it has outstanding
773     events but its callback has not yet been invoked). As long as a watcher
774     is pending (but not active) you must not call an init function on it (but
775 root 1.70 <code>ev_TYPE_set</code> is safe), you must not change its priority, and you must
776     make sure the watcher is available to libev (e.g. you cannot <code>free ()</code>
777     it).</p>
778 root 1.37 </dd>
779 root 1.56 <dt>callback ev_cb (ev_TYPE *watcher)</dt>
780 root 1.37 <dd>
781     <p>Returns the callback currently set on the watcher.</p>
782     </dd>
783     <dt>ev_cb_set (ev_TYPE *watcher, callback)</dt>
784     <dd>
785     <p>Change the callback. You can change the callback at virtually any time
786     (modulo threads).</p>
787     </dd>
788 root 1.64 <dt>ev_set_priority (ev_TYPE *watcher, priority)</dt>
789     <dt>int ev_priority (ev_TYPE *watcher)</dt>
790     <dd>
791     <p>Set and query the priority of the watcher. The priority is a small
792     integer between <code>EV_MAXPRI</code> (default: <code>2</code>) and <code>EV_MINPRI</code>
793     (default: <code>-2</code>). Pending watchers with higher priority will be invoked
794     before watchers with lower priority, but priority will not keep watchers
795     from being executed (except for <code>ev_idle</code> watchers).</p>
796     <p>This means that priorities are <i>only</i> used for ordering callback
797     invocation after new events have been received. This is useful, for
798     example, to reduce latency after idling, or more often, to bind two
799     watchers on the same event and make sure one is called first.</p>
800     <p>If you need to suppress invocation when higher priority events are pending
801     you need to look at <code>ev_idle</code> watchers, which provide this functionality.</p>
802 root 1.70 <p>You <i>must not</i> change the priority of a watcher as long as it is active or
803     pending.</p>
804 root 1.64 <p>The default priority used by watchers when no priority has been set is
805     always <code>0</code>, which is supposed to not be too high and not be too low :).</p>
806     <p>Setting a priority outside the range of <code>EV_MINPRI</code> to <code>EV_MAXPRI</code> is
807     fine, as long as you do not mind that the priority value you query might
808     or might not have been adjusted to be within valid range.</p>
809     </dd>
810 root 1.70 <dt>ev_invoke (loop, ev_TYPE *watcher, int revents)</dt>
811     <dd>
812     <p>Invoke the <code>watcher</code> with the given <code>loop</code> and <code>revents</code>. Neither
813     <code>loop</code> nor <code>revents</code> need to be valid as long as the watcher callback
814     can deal with that fact.</p>
815     </dd>
816     <dt>int ev_clear_pending (loop, ev_TYPE *watcher)</dt>
817     <dd>
818     <p>If the watcher is pending, this function returns clears its pending status
819     and returns its <code>revents</code> bitset (as if its callback was invoked). If the
820     watcher isn't pending it does nothing and returns <code>0</code>.</p>
821     </dd>
822 root 1.37 </dl>
823    
824    
825    
826    
827    
828     </div>
829 root 1.1 <h2 id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH">ASSOCIATING CUSTOM DATA WITH A WATCHER</h2>
830     <div id="ASSOCIATING_CUSTOM_DATA_WITH_A_WATCH-2">
831     <p>Each watcher has, by default, a member <code>void *data</code> that you can change
832 root 1.14 and read at any time, libev will completely ignore it. This can be used
833 root 1.1 to associate arbitrary data with your watcher. If you need more data and
834     don't want to allocate memory and store a pointer to it in that data
835     member, you can also &quot;subclass&quot; the watcher type and provide your own
836     data:</p>
837     <pre> struct my_io
838     {
839     struct ev_io io;
840     int otherfd;
841     void *somedata;
842     struct whatever *mostinteresting;
843     }
844    
845     </pre>
846     <p>And since your callback will be called with a pointer to the watcher, you
847     can cast it back to your own type:</p>
848     <pre> static void my_cb (struct ev_loop *loop, struct ev_io *w_, int revents)
849     {
850     struct my_io *w = (struct my_io *)w_;
851     ...
852     }
853    
854     </pre>
855 root 1.56 <p>More interesting and less C-conformant ways of casting your callback type
856     instead have been omitted.</p>
857     <p>Another common scenario is having some data structure with multiple
858     watchers:</p>
859     <pre> struct my_biggy
860     {
861     int some_data;
862     ev_timer t1;
863     ev_timer t2;
864     }
865 root 1.1
866 root 1.56 </pre>
867     <p>In this case getting the pointer to <code>my_biggy</code> is a bit more complicated,
868     you need to use <code>offsetof</code>:</p>
869     <pre> #include &lt;stddef.h&gt;
870 root 1.1
871 root 1.56 static void
872     t1_cb (EV_P_ struct ev_timer *w, int revents)
873     {
874     struct my_biggy big = (struct my_biggy *
875     (((char *)w) - offsetof (struct my_biggy, t1));
876     }
877    
878     static void
879     t2_cb (EV_P_ struct ev_timer *w, int revents)
880     {
881     struct my_biggy big = (struct my_biggy *
882     (((char *)w) - offsetof (struct my_biggy, t2));
883     }
884 root 1.1
885    
886    
887 root 1.56
888     </pre>
889    
890 root 1.1 </div>
891 root 1.55 <h1 id="WATCHER_TYPES">WATCHER TYPES</h1>
892 root 1.1 <div id="WATCHER_TYPES_CONTENT">
893     <p>This section describes each watcher in detail, but will not repeat
894 root 1.48 information given in the last section. Any initialisation/set macros,
895     functions and members specific to the watcher type are explained.</p>
896     <p>Members are additionally marked with either <i>[read-only]</i>, meaning that,
897     while the watcher is active, you can look at the member and expect some
898     sensible content, but you must not modify it (you can modify it while the
899     watcher is stopped to your hearts content), or <i>[read-write]</i>, which
900     means you can expect it to have some sensible content while the watcher
901     is active, but you can also modify it. Modifying it may not do something
902     sensible or take immediate effect (or do anything at all), but libev will
903     not crash or malfunction in any way.</p>
904 root 1.1
905 root 1.35
906    
907    
908    
909 root 1.1 </div>
910 root 1.43 <h2 id="code_ev_io_code_is_this_file_descrip"><code>ev_io</code> - is this file descriptor readable or writable?</h2>
911 root 1.11 <div id="code_ev_io_code_is_this_file_descrip-2">
912 root 1.4 <p>I/O watchers check whether a file descriptor is readable or writable
913 root 1.43 in each iteration of the event loop, or, more precisely, when reading
914     would not block the process and writing would at least be able to write
915     some data. This behaviour is called level-triggering because you keep
916     receiving events as long as the condition persists. Remember you can stop
917     the watcher if you don't want to act on the event and neither want to
918     receive future events.</p>
919 root 1.25 <p>In general you can register as many read and/or write event watchers per
920 root 1.8 fd as you want (as long as you don't confuse yourself). Setting all file
921     descriptors to non-blocking mode is also usually a good idea (but not
922     required if you know what you are doing).</p>
923     <p>You have to be careful with dup'ed file descriptors, though. Some backends
924     (the linux epoll backend is a notable example) cannot handle dup'ed file
925     descriptors correctly if you register interest in two or more fds pointing
926 root 1.43 to the same underlying file/socket/etc. description (that is, they share
927 root 1.26 the same underlying &quot;file open&quot;).</p>
928 root 1.8 <p>If you must do this, then force the use of a known-to-be-good backend
929 root 1.32 (at the time of this writing, this includes only <code>EVBACKEND_SELECT</code> and
930     <code>EVBACKEND_POLL</code>).</p>
931 root 1.43 <p>Another thing you have to watch out for is that it is quite easy to
932     receive &quot;spurious&quot; readyness notifications, that is your callback might
933     be called with <code>EV_READ</code> but a subsequent <code>read</code>(2) will actually block
934     because there is no data. Not only are some backends known to create a
935     lot of those (for example solaris ports), it is very easy to get into
936     this situation even with a relatively standard program structure. Thus
937     it is best to always use non-blocking I/O: An extra <code>read</code>(2) returning
938     <code>EAGAIN</code> is far preferable to a program hanging until some data arrives.</p>
939     <p>If you cannot run the fd in non-blocking mode (for example you should not
940     play around with an Xlib connection), then you have to seperately re-test
941 root 1.65 whether a file descriptor is really ready with a known-to-be good interface
942 root 1.43 such as poll (fortunately in our Xlib example, Xlib already does this on
943     its own, so its quite safe to use).</p>
944 root 1.1 <dl>
945     <dt>ev_io_init (ev_io *, callback, int fd, int events)</dt>
946     <dt>ev_io_set (ev_io *, int fd, int events)</dt>
947     <dd>
948 root 1.43 <p>Configures an <code>ev_io</code> watcher. The <code>fd</code> is the file descriptor to
949     rceeive events for and events is either <code>EV_READ</code>, <code>EV_WRITE</code> or
950     <code>EV_READ | EV_WRITE</code> to receive the given events.</p>
951 root 1.1 </dd>
952 root 1.48 <dt>int fd [read-only]</dt>
953     <dd>
954     <p>The file descriptor being watched.</p>
955     </dd>
956     <dt>int events [read-only]</dt>
957     <dd>
958     <p>The events being watched.</p>
959     </dd>
960 root 1.1 </dl>
961 root 1.54 <p>Example: Call <code>stdin_readable_cb</code> when STDIN_FILENO has become, well
962 root 1.35 readable, but only once. Since it is likely line-buffered, you could
963 root 1.54 attempt to read a whole line in the callback.</p>
964 root 1.35 <pre> static void
965     stdin_readable_cb (struct ev_loop *loop, struct ev_io *w, int revents)
966     {
967     ev_io_stop (loop, w);
968     .. read from stdin here (or from w-&gt;fd) and haqndle any I/O errors
969     }
970    
971     ...
972     struct ev_loop *loop = ev_default_init (0);
973     struct ev_io stdin_readable;
974     ev_io_init (&amp;stdin_readable, stdin_readable_cb, STDIN_FILENO, EV_READ);
975     ev_io_start (loop, &amp;stdin_readable);
976     ev_loop (loop, 0);
977    
978    
979    
980    
981     </pre>
982 root 1.1
983     </div>
984 root 1.43 <h2 id="code_ev_timer_code_relative_and_opti"><code>ev_timer</code> - relative and optionally repeating timeouts</h2>
985 root 1.10 <div id="code_ev_timer_code_relative_and_opti-2">
986 root 1.1 <p>Timer watchers are simple relative timers that generate an event after a
987     given time, and optionally repeating in regular intervals after that.</p>
988     <p>The timers are based on real time, that is, if you register an event that
989 root 1.25 times out after an hour and you reset your system clock to last years
990 root 1.1 time, it will still time out after (roughly) and hour. &quot;Roughly&quot; because
991 root 1.28 detecting time jumps is hard, and some inaccuracies are unavoidable (the
992 root 1.1 monotonic clock option helps a lot here).</p>
993 root 1.9 <p>The relative timeouts are calculated relative to the <code>ev_now ()</code>
994     time. This is usually the right thing as this timestamp refers to the time
995 root 1.28 of the event triggering whatever timeout you are modifying/starting. If
996     you suspect event processing to be delayed and you <i>need</i> to base the timeout
997 root 1.25 on the current time, use something like this to adjust for this:</p>
998 root 1.9 <pre> ev_timer_set (&amp;timer, after + ev_now () - ev_time (), 0.);
999    
1000     </pre>
1001 root 1.28 <p>The callback is guarenteed to be invoked only when its timeout has passed,
1002     but if multiple timers become ready during the same loop iteration then
1003     order of execution is undefined.</p>
1004 root 1.1 <dl>
1005     <dt>ev_timer_init (ev_timer *, callback, ev_tstamp after, ev_tstamp repeat)</dt>
1006     <dt>ev_timer_set (ev_timer *, ev_tstamp after, ev_tstamp repeat)</dt>
1007     <dd>
1008     <p>Configure the timer to trigger after <code>after</code> seconds. If <code>repeat</code> is
1009     <code>0.</code>, then it will automatically be stopped. If it is positive, then the
1010     timer will automatically be configured to trigger again <code>repeat</code> seconds
1011     later, again, and again, until stopped manually.</p>
1012     <p>The timer itself will do a best-effort at avoiding drift, that is, if you
1013     configure a timer to trigger every 10 seconds, then it will trigger at
1014     exactly 10 second intervals. If, however, your program cannot keep up with
1015 root 1.25 the timer (because it takes longer than those 10 seconds to do stuff) the
1016 root 1.1 timer will not fire more than once per event loop iteration.</p>
1017     </dd>
1018     <dt>ev_timer_again (loop)</dt>
1019     <dd>
1020     <p>This will act as if the timer timed out and restart it again if it is
1021     repeating. The exact semantics are:</p>
1022 root 1.61 <p>If the timer is pending, its pending status is cleared.</p>
1023     <p>If the timer is started but nonrepeating, stop it (as if it timed out).</p>
1024     <p>If the timer is repeating, either start it if necessary (with the
1025     <code>repeat</code> value), or reset the running timer to the <code>repeat</code> value.</p>
1026 root 1.1 <p>This sounds a bit complicated, but here is a useful and typical
1027 root 1.61 example: Imagine you have a tcp connection and you want a so-called idle
1028     timeout, that is, you want to be called when there have been, say, 60
1029     seconds of inactivity on the socket. The easiest way to do this is to
1030     configure an <code>ev_timer</code> with a <code>repeat</code> value of <code>60</code> and then call
1031 root 1.48 <code>ev_timer_again</code> each time you successfully read or write some data. If
1032     you go into an idle state where you do not expect data to travel on the
1033 root 1.61 socket, you can <code>ev_timer_stop</code> the timer, and <code>ev_timer_again</code> will
1034     automatically restart it if need be.</p>
1035     <p>That means you can ignore the <code>after</code> value and <code>ev_timer_start</code>
1036     altogether and only ever use the <code>repeat</code> value and <code>ev_timer_again</code>:</p>
1037 root 1.48 <pre> ev_timer_init (timer, callback, 0., 5.);
1038     ev_timer_again (loop, timer);
1039     ...
1040     timer-&gt;again = 17.;
1041     ev_timer_again (loop, timer);
1042     ...
1043     timer-&gt;again = 10.;
1044     ev_timer_again (loop, timer);
1045    
1046     </pre>
1047 root 1.61 <p>This is more slightly efficient then stopping/starting the timer each time
1048     you want to modify its timeout value.</p>
1049 root 1.48 </dd>
1050     <dt>ev_tstamp repeat [read-write]</dt>
1051     <dd>
1052     <p>The current <code>repeat</code> value. Will be used each time the watcher times out
1053     or <code>ev_timer_again</code> is called and determines the next timeout (if any),
1054     which is also when any modifications are taken into account.</p>
1055 root 1.1 </dd>
1056     </dl>
1057 root 1.54 <p>Example: Create a timer that fires after 60 seconds.</p>
1058 root 1.35 <pre> static void
1059     one_minute_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1060     {
1061     .. one minute over, w is actually stopped right here
1062     }
1063    
1064     struct ev_timer mytimer;
1065     ev_timer_init (&amp;mytimer, one_minute_cb, 60., 0.);
1066     ev_timer_start (loop, &amp;mytimer);
1067    
1068     </pre>
1069 root 1.54 <p>Example: Create a timeout timer that times out after 10 seconds of
1070 root 1.35 inactivity.</p>
1071     <pre> static void
1072     timeout_cb (struct ev_loop *loop, struct ev_timer *w, int revents)
1073     {
1074     .. ten seconds without any activity
1075     }
1076    
1077     struct ev_timer mytimer;
1078     ev_timer_init (&amp;mytimer, timeout_cb, 0., 10.); /* note, only repeat used */
1079     ev_timer_again (&amp;mytimer); /* start timer */
1080     ev_loop (loop, 0);
1081    
1082     // and in some piece of code that gets executed on any &quot;activity&quot;:
1083     // reset the timeout to start ticking again at 10 seconds
1084     ev_timer_again (&amp;mytimer);
1085    
1086    
1087    
1088    
1089     </pre>
1090 root 1.1
1091     </div>
1092 root 1.43 <h2 id="code_ev_periodic_code_to_cron_or_not"><code>ev_periodic</code> - to cron or not to cron?</h2>
1093 root 1.10 <div id="code_ev_periodic_code_to_cron_or_not-2">
1094 root 1.1 <p>Periodic watchers are also timers of a kind, but they are very versatile
1095     (and unfortunately a bit complex).</p>
1096 root 1.10 <p>Unlike <code>ev_timer</code>'s, they are not based on real time (or relative time)
1097 root 1.1 but on wallclock time (absolute time). You can tell a periodic watcher
1098     to trigger &quot;at&quot; some specific point in time. For example, if you tell a
1099 root 1.39 periodic watcher to trigger in 10 seconds (by specifiying e.g. <code>ev_now ()
1100     + 10.</code>) and then reset your system clock to the last year, then it will
1101 root 1.10 take a year to trigger the event (unlike an <code>ev_timer</code>, which would trigger
1102 root 1.74 roughly 10 seconds later).</p>
1103 root 1.1 <p>They can also be used to implement vastly more complex timers, such as
1104 root 1.74 triggering an event on each midnight, local time or other, complicated,
1105     rules.</p>
1106 root 1.28 <p>As with timers, the callback is guarenteed to be invoked only when the
1107     time (<code>at</code>) has been passed, but if multiple periodic timers become ready
1108     during the same loop iteration then order of execution is undefined.</p>
1109 root 1.1 <dl>
1110     <dt>ev_periodic_init (ev_periodic *, callback, ev_tstamp at, ev_tstamp interval, reschedule_cb)</dt>
1111     <dt>ev_periodic_set (ev_periodic *, ev_tstamp after, ev_tstamp repeat, reschedule_cb)</dt>
1112     <dd>
1113     <p>Lots of arguments, lets sort it out... There are basically three modes of
1114     operation, and we will explain them from simplest to complex:</p>
1115     <p>
1116     <dl>
1117 root 1.74 <dt>* absolute timer (at = time, interval = reschedule_cb = 0)</dt>
1118 root 1.1 <dd>
1119     <p>In this configuration the watcher triggers an event at the wallclock time
1120     <code>at</code> and doesn't repeat. It will not adjust when a time jump occurs,
1121     that is, if it is to be run at January 1st 2011 then it will run when the
1122     system time reaches or surpasses this time.</p>
1123     </dd>
1124 root 1.74 <dt>* non-repeating interval timer (at = offset, interval &gt; 0, reschedule_cb = 0)</dt>
1125 root 1.1 <dd>
1126     <p>In this mode the watcher will always be scheduled to time out at the next
1127 root 1.74 <code>at + N * interval</code> time (for some integer N, which can also be negative)
1128     and then repeat, regardless of any time jumps.</p>
1129 root 1.1 <p>This can be used to create timers that do not drift with respect to system
1130     time:</p>
1131     <pre> ev_periodic_set (&amp;periodic, 0., 3600., 0);
1132    
1133     </pre>
1134     <p>This doesn't mean there will always be 3600 seconds in between triggers,
1135     but only that the the callback will be called when the system time shows a
1136 root 1.12 full hour (UTC), or more correctly, when the system time is evenly divisible
1137 root 1.1 by 3600.</p>
1138     <p>Another way to think about it (for the mathematically inclined) is that
1139 root 1.10 <code>ev_periodic</code> will try to run the callback in this mode at the next possible
1140 root 1.1 time where <code>time = at (mod interval)</code>, regardless of any time jumps.</p>
1141 root 1.74 <p>For numerical stability it is preferable that the <code>at</code> value is near
1142     <code>ev_now ()</code> (the current time), but there is no range requirement for
1143     this value.</p>
1144 root 1.1 </dd>
1145 root 1.74 <dt>* manual reschedule mode (at and interval ignored, reschedule_cb = callback)</dt>
1146 root 1.1 <dd>
1147     <p>In this mode the values for <code>interval</code> and <code>at</code> are both being
1148     ignored. Instead, each time the periodic watcher gets scheduled, the
1149     reschedule callback will be called with the watcher as first, and the
1150     current time as second argument.</p>
1151 root 1.21 <p>NOTE: <i>This callback MUST NOT stop or destroy any periodic watcher,
1152     ever, or make any event loop modifications</i>. If you need to stop it,
1153     return <code>now + 1e30</code> (or so, fudge fudge) and stop it afterwards (e.g. by
1154 root 1.74 starting an <code>ev_prepare</code> watcher, which is legal).</p>
1155 root 1.13 <p>Its prototype is <code>ev_tstamp (*reschedule_cb)(struct ev_periodic *w,
1156     ev_tstamp now)</code>, e.g.:</p>
1157 root 1.1 <pre> static ev_tstamp my_rescheduler (struct ev_periodic *w, ev_tstamp now)
1158     {
1159     return now + 60.;
1160     }
1161    
1162     </pre>
1163     <p>It must return the next time to trigger, based on the passed time value
1164     (that is, the lowest time value larger than to the second argument). It
1165     will usually be called just before the callback will be triggered, but
1166     might be called at other times, too.</p>
1167 root 1.21 <p>NOTE: <i>This callback must always return a time that is later than the
1168 root 1.22 passed <code>now</code> value</i>. Not even <code>now</code> itself will do, it <i>must</i> be larger.</p>
1169 root 1.1 <p>This can be used to create very complex timers, such as a timer that
1170     triggers on each midnight, local time. To do this, you would calculate the
1171 root 1.22 next midnight after <code>now</code> and return the timestamp value for this. How
1172     you do this is, again, up to you (but it is not trivial, which is the main
1173     reason I omitted it as an example).</p>
1174 root 1.1 </dd>
1175     </dl>
1176     </p>
1177     </dd>
1178     <dt>ev_periodic_again (loop, ev_periodic *)</dt>
1179     <dd>
1180     <p>Simply stops and restarts the periodic watcher again. This is only useful
1181     when you changed some parameters or the reschedule callback would return
1182     a different time than the last time it was called (e.g. in a crond like
1183     program when the crontabs have changed).</p>
1184     </dd>
1185 root 1.74 <dt>ev_tstamp offset [read-write]</dt>
1186     <dd>
1187     <p>When repeating, this contains the offset value, otherwise this is the
1188     absolute point in time (the <code>at</code> value passed to <code>ev_periodic_set</code>).</p>
1189     <p>Can be modified any time, but changes only take effect when the periodic
1190     timer fires or <code>ev_periodic_again</code> is being called.</p>
1191     </dd>
1192 root 1.48 <dt>ev_tstamp interval [read-write]</dt>
1193     <dd>
1194     <p>The current interval value. Can be modified any time, but changes only
1195     take effect when the periodic timer fires or <code>ev_periodic_again</code> is being
1196     called.</p>
1197     </dd>
1198     <dt>ev_tstamp (*reschedule_cb)(struct ev_periodic *w, ev_tstamp now) [read-write]</dt>
1199     <dd>
1200     <p>The current reschedule callback, or <code>0</code>, if this functionality is
1201     switched off. Can be changed any time, but changes only take effect when
1202     the periodic timer fires or <code>ev_periodic_again</code> is being called.</p>
1203     </dd>
1204 root 1.1 </dl>
1205 root 1.54 <p>Example: Call a callback every hour, or, more precisely, whenever the
1206 root 1.35 system clock is divisible by 3600. The callback invocation times have
1207     potentially a lot of jittering, but good long-term stability.</p>
1208     <pre> static void
1209     clock_cb (struct ev_loop *loop, struct ev_io *w, int revents)
1210     {
1211     ... its now a full hour (UTC, or TAI or whatever your clock follows)
1212     }
1213    
1214     struct ev_periodic hourly_tick;
1215     ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 3600., 0);
1216     ev_periodic_start (loop, &amp;hourly_tick);
1217    
1218     </pre>
1219 root 1.54 <p>Example: The same as above, but use a reschedule callback to do it:</p>
1220 root 1.35 <pre> #include &lt;math.h&gt;
1221    
1222     static ev_tstamp
1223     my_scheduler_cb (struct ev_periodic *w, ev_tstamp now)
1224     {
1225     return fmod (now, 3600.) + 3600.;
1226     }
1227    
1228     ev_periodic_init (&amp;hourly_tick, clock_cb, 0., 0., my_scheduler_cb);
1229    
1230     </pre>
1231 root 1.54 <p>Example: Call a callback every hour, starting now:</p>
1232 root 1.35 <pre> struct ev_periodic hourly_tick;
1233     ev_periodic_init (&amp;hourly_tick, clock_cb,
1234     fmod (ev_now (loop), 3600.), 3600., 0);
1235     ev_periodic_start (loop, &amp;hourly_tick);
1236    
1237    
1238    
1239    
1240     </pre>
1241 root 1.1
1242     </div>
1243 root 1.43 <h2 id="code_ev_signal_code_signal_me_when_a"><code>ev_signal</code> - signal me when a signal gets signalled!</h2>
1244 root 1.10 <div id="code_ev_signal_code_signal_me_when_a-2">
1245 root 1.1 <p>Signal watchers will trigger an event when the process receives a specific
1246     signal one or more times. Even though signals are very asynchronous, libev
1247 root 1.9 will try it's best to deliver signals synchronously, i.e. as part of the
1248 root 1.1 normal event processing, like any other event.</p>
1249 root 1.14 <p>You can configure as many watchers as you like per signal. Only when the
1250 root 1.1 first watcher gets started will libev actually register a signal watcher
1251     with the kernel (thus it coexists with your own signal handlers as long
1252     as you don't register any with libev). Similarly, when the last signal
1253     watcher for a signal is stopped libev will reset the signal handler to
1254     SIG_DFL (regardless of what it was set to before).</p>
1255     <dl>
1256     <dt>ev_signal_init (ev_signal *, callback, int signum)</dt>
1257     <dt>ev_signal_set (ev_signal *, int signum)</dt>
1258     <dd>
1259     <p>Configures the watcher to trigger on the given signal number (usually one
1260     of the <code>SIGxxx</code> constants).</p>
1261     </dd>
1262 root 1.48 <dt>int signum [read-only]</dt>
1263     <dd>
1264     <p>The signal the watcher watches out for.</p>
1265     </dd>
1266 root 1.1 </dl>
1267    
1268 root 1.36
1269    
1270    
1271    
1272 root 1.1 </div>
1273 root 1.43 <h2 id="code_ev_child_code_watch_out_for_pro"><code>ev_child</code> - watch out for process status changes</h2>
1274     <div id="code_ev_child_code_watch_out_for_pro-2">
1275 root 1.1 <p>Child watchers trigger when your process receives a SIGCHLD in response to
1276     some child status changes (most typically when a child of yours dies).</p>
1277     <dl>
1278     <dt>ev_child_init (ev_child *, callback, int pid)</dt>
1279     <dt>ev_child_set (ev_child *, int pid)</dt>
1280     <dd>
1281     <p>Configures the watcher to wait for status changes of process <code>pid</code> (or
1282     <i>any</i> process if <code>pid</code> is specified as <code>0</code>). The callback can look
1283     at the <code>rstatus</code> member of the <code>ev_child</code> watcher structure to see
1284 root 1.14 the status word (use the macros from <code>sys/wait.h</code> and see your systems
1285     <code>waitpid</code> documentation). The <code>rpid</code> member contains the pid of the
1286     process causing the status change.</p>
1287 root 1.1 </dd>
1288 root 1.48 <dt>int pid [read-only]</dt>
1289     <dd>
1290     <p>The process id this watcher watches out for, or <code>0</code>, meaning any process id.</p>
1291     </dd>
1292     <dt>int rpid [read-write]</dt>
1293     <dd>
1294     <p>The process id that detected a status change.</p>
1295     </dd>
1296     <dt>int rstatus [read-write]</dt>
1297     <dd>
1298     <p>The process exit/trace status caused by <code>rpid</code> (see your systems
1299     <code>waitpid</code> and <code>sys/wait.h</code> documentation for details).</p>
1300     </dd>
1301 root 1.1 </dl>
1302 root 1.54 <p>Example: Try to exit cleanly on SIGINT and SIGTERM.</p>
1303 root 1.35 <pre> static void
1304     sigint_cb (struct ev_loop *loop, struct ev_signal *w, int revents)
1305     {
1306     ev_unloop (loop, EVUNLOOP_ALL);
1307     }
1308    
1309     struct ev_signal signal_watcher;
1310     ev_signal_init (&amp;signal_watcher, sigint_cb, SIGINT);
1311     ev_signal_start (loop, &amp;sigint_cb);
1312    
1313    
1314    
1315    
1316     </pre>
1317 root 1.1
1318     </div>
1319 root 1.48 <h2 id="code_ev_stat_code_did_the_file_attri"><code>ev_stat</code> - did the file attributes just change?</h2>
1320     <div id="code_ev_stat_code_did_the_file_attri-2">
1321     <p>This watches a filesystem path for attribute changes. That is, it calls
1322     <code>stat</code> regularly (or when the OS says it changed) and sees if it changed
1323     compared to the last time, invoking the callback if it did.</p>
1324     <p>The path does not need to exist: changing from &quot;path exists&quot; to &quot;path does
1325     not exist&quot; is a status change like any other. The condition &quot;path does
1326     not exist&quot; is signified by the <code>st_nlink</code> field being zero (which is
1327     otherwise always forced to be at least one) and all the other fields of
1328     the stat buffer having unspecified contents.</p>
1329 root 1.60 <p>The path <i>should</i> be absolute and <i>must not</i> end in a slash. If it is
1330     relative and your working directory changes, the behaviour is undefined.</p>
1331 root 1.48 <p>Since there is no standard to do this, the portable implementation simply
1332 root 1.57 calls <code>stat (2)</code> regularly on the path to see if it changed somehow. You
1333 root 1.48 can specify a recommended polling interval for this case. If you specify
1334     a polling interval of <code>0</code> (highly recommended!) then a <i>suitable,
1335     unspecified default</i> value will be used (which you can expect to be around
1336     five seconds, although this might change dynamically). Libev will also
1337     impose a minimum interval which is currently around <code>0.1</code>, but thats
1338     usually overkill.</p>
1339     <p>This watcher type is not meant for massive numbers of stat watchers,
1340     as even with OS-supported change notifications, this can be
1341     resource-intensive.</p>
1342 root 1.57 <p>At the time of this writing, only the Linux inotify interface is
1343     implemented (implementing kqueue support is left as an exercise for the
1344     reader). Inotify will be used to give hints only and should not change the
1345     semantics of <code>ev_stat</code> watchers, which means that libev sometimes needs
1346     to fall back to regular polling again even with inotify, but changes are
1347     usually detected immediately, and if the file exists there will be no
1348     polling.</p>
1349 root 1.48 <dl>
1350     <dt>ev_stat_init (ev_stat *, callback, const char *path, ev_tstamp interval)</dt>
1351     <dt>ev_stat_set (ev_stat *, const char *path, ev_tstamp interval)</dt>
1352     <dd>
1353     <p>Configures the watcher to wait for status changes of the given
1354     <code>path</code>. The <code>interval</code> is a hint on how quickly a change is expected to
1355     be detected and should normally be specified as <code>0</code> to let libev choose
1356     a suitable value. The memory pointed to by <code>path</code> must point to the same
1357     path for as long as the watcher is active.</p>
1358     <p>The callback will be receive <code>EV_STAT</code> when a change was detected,
1359     relative to the attributes at the time the watcher was started (or the
1360     last change was detected).</p>
1361     </dd>
1362     <dt>ev_stat_stat (ev_stat *)</dt>
1363     <dd>
1364     <p>Updates the stat buffer immediately with new values. If you change the
1365     watched path in your callback, you could call this fucntion to avoid
1366     detecting this change (while introducing a race condition). Can also be
1367     useful simply to find out the new values.</p>
1368     </dd>
1369     <dt>ev_statdata attr [read-only]</dt>
1370     <dd>
1371     <p>The most-recently detected attributes of the file. Although the type is of
1372     <code>ev_statdata</code>, this is usually the (or one of the) <code>struct stat</code> types
1373     suitable for your system. If the <code>st_nlink</code> member is <code>0</code>, then there
1374     was some error while <code>stat</code>ing the file.</p>
1375     </dd>
1376     <dt>ev_statdata prev [read-only]</dt>
1377     <dd>
1378     <p>The previous attributes of the file. The callback gets invoked whenever
1379     <code>prev</code> != <code>attr</code>.</p>
1380     </dd>
1381     <dt>ev_tstamp interval [read-only]</dt>
1382     <dd>
1383     <p>The specified interval.</p>
1384     </dd>
1385     <dt>const char *path [read-only]</dt>
1386     <dd>
1387     <p>The filesystem path that is being watched.</p>
1388     </dd>
1389     </dl>
1390     <p>Example: Watch <code>/etc/passwd</code> for attribute changes.</p>
1391     <pre> static void
1392     passwd_cb (struct ev_loop *loop, ev_stat *w, int revents)
1393     {
1394     /* /etc/passwd changed in some way */
1395     if (w-&gt;attr.st_nlink)
1396     {
1397     printf (&quot;passwd current size %ld\n&quot;, (long)w-&gt;attr.st_size);
1398     printf (&quot;passwd current atime %ld\n&quot;, (long)w-&gt;attr.st_mtime);
1399     printf (&quot;passwd current mtime %ld\n&quot;, (long)w-&gt;attr.st_mtime);
1400     }
1401     else
1402     /* you shalt not abuse printf for puts */
1403     puts (&quot;wow, /etc/passwd is not there, expect problems. &quot;
1404     &quot;if this is windows, they already arrived\n&quot;);
1405     }
1406    
1407     ...
1408     ev_stat passwd;
1409    
1410     ev_stat_init (&amp;passwd, passwd_cb, &quot;/etc/passwd&quot;);
1411     ev_stat_start (loop, &amp;passwd);
1412    
1413    
1414    
1415    
1416     </pre>
1417    
1418     </div>
1419 root 1.43 <h2 id="code_ev_idle_code_when_you_ve_got_no"><code>ev_idle</code> - when you've got nothing better to do...</h2>
1420 root 1.10 <div id="code_ev_idle_code_when_you_ve_got_no-2">
1421 root 1.64 <p>Idle watchers trigger events when no other events of the same or higher
1422     priority are pending (prepare, check and other idle watchers do not
1423     count).</p>
1424     <p>That is, as long as your process is busy handling sockets or timeouts
1425     (or even signals, imagine) of the same or higher priority it will not be
1426     triggered. But when your process is idle (or only lower-priority watchers
1427     are pending), the idle watchers are being called once per event loop
1428     iteration - until stopped, that is, or your process receives more events
1429     and becomes busy again with higher priority stuff.</p>
1430 root 1.1 <p>The most noteworthy effect is that as long as any idle watchers are
1431     active, the process will not block when waiting for new events.</p>
1432     <p>Apart from keeping your process non-blocking (which is a useful
1433     effect on its own sometimes), idle watchers are a good place to do
1434     &quot;pseudo-background processing&quot;, or delay processing stuff to after the
1435     event loop has handled all outstanding events.</p>
1436     <dl>
1437     <dt>ev_idle_init (ev_signal *, callback)</dt>
1438     <dd>
1439     <p>Initialises and configures the idle watcher - it has no parameters of any
1440     kind. There is a <code>ev_idle_set</code> macro, but using it is utterly pointless,
1441     believe me.</p>
1442     </dd>
1443     </dl>
1444 root 1.54 <p>Example: Dynamically allocate an <code>ev_idle</code> watcher, start it, and in the
1445     callback, free it. Also, use no error checking, as usual.</p>
1446 root 1.35 <pre> static void
1447     idle_cb (struct ev_loop *loop, struct ev_idle *w, int revents)
1448     {
1449     free (w);
1450     // now do something you wanted to do when the program has
1451     // no longer asnything immediate to do.
1452     }
1453    
1454     struct ev_idle *idle_watcher = malloc (sizeof (struct ev_idle));
1455     ev_idle_init (idle_watcher, idle_cb);
1456     ev_idle_start (loop, idle_cb);
1457    
1458    
1459    
1460    
1461     </pre>
1462 root 1.1
1463     </div>
1464 root 1.43 <h2 id="code_ev_prepare_code_and_code_ev_che"><code>ev_prepare</code> and <code>ev_check</code> - customise your event loop!</h2>
1465 root 1.16 <div id="code_ev_prepare_code_and_code_ev_che-2">
1466 root 1.14 <p>Prepare and check watchers are usually (but not always) used in tandem:
1467 root 1.23 prepare watchers get invoked before the process blocks and check watchers
1468 root 1.14 afterwards.</p>
1469 root 1.46 <p>You <i>must not</i> call <code>ev_loop</code> or similar functions that enter
1470     the current event loop from either <code>ev_prepare</code> or <code>ev_check</code>
1471     watchers. Other loops than the current one are fine, however. The
1472     rationale behind this is that you do not need to check for recursion in
1473     those watchers, i.e. the sequence will always be <code>ev_prepare</code>, blocking,
1474     <code>ev_check</code> so if you have one watcher of each kind they will always be
1475     called in pairs bracketing the blocking call.</p>
1476 root 1.36 <p>Their main purpose is to integrate other event mechanisms into libev and
1477     their use is somewhat advanced. This could be used, for example, to track
1478     variable changes, implement your own watchers, integrate net-snmp or a
1479 root 1.46 coroutine library and lots more. They are also occasionally useful if
1480     you cache some data and want to flush it before blocking (for example,
1481     in X programs you might want to do an <code>XFlush ()</code> in an <code>ev_prepare</code>
1482     watcher).</p>
1483 root 1.1 <p>This is done by examining in each prepare call which file descriptors need
1484 root 1.14 to be watched by the other library, registering <code>ev_io</code> watchers for
1485     them and starting an <code>ev_timer</code> watcher for any timeouts (many libraries
1486     provide just this functionality). Then, in the check watcher you check for
1487     any events that occured (by checking the pending status of all watchers
1488     and stopping them) and call back into the library. The I/O and timer
1489 root 1.23 callbacks will never actually be called (but must be valid nevertheless,
1490 root 1.14 because you never know, you know?).</p>
1491     <p>As another example, the Perl Coro module uses these hooks to integrate
1492 root 1.1 coroutines into libev programs, by yielding to other active coroutines
1493     during each prepare and only letting the process block if no coroutines
1494 root 1.23 are ready to run (it's actually more complicated: it only runs coroutines
1495     with priority higher than or equal to the event loop and one coroutine
1496     of lower priority, but only once, using idle watchers to keep the event
1497     loop from blocking if lower-priority coroutines are active, thus mapping
1498     low-priority coroutines to idle/background tasks).</p>
1499 root 1.73 <p>It is recommended to give <code>ev_check</code> watchers highest (<code>EV_MAXPRI</code>)
1500     priority, to ensure that they are being run before any other watchers
1501     after the poll. Also, <code>ev_check</code> watchers (and <code>ev_prepare</code> watchers,
1502     too) should not activate (&quot;feed&quot;) events into libev. While libev fully
1503     supports this, they will be called before other <code>ev_check</code> watchers did
1504     their job. As <code>ev_check</code> watchers are often used to embed other event
1505     loops those other event loops might be in an unusable state until their
1506     <code>ev_check</code> watcher ran (always remind yourself to coexist peacefully with
1507     others).</p>
1508 root 1.1 <dl>
1509     <dt>ev_prepare_init (ev_prepare *, callback)</dt>
1510     <dt>ev_check_init (ev_check *, callback)</dt>
1511     <dd>
1512     <p>Initialises and configures the prepare or check watcher - they have no
1513     parameters of any kind. There are <code>ev_prepare_set</code> and <code>ev_check_set</code>
1514 root 1.14 macros, but using them is utterly, utterly and completely pointless.</p>
1515 root 1.1 </dd>
1516     </dl>
1517 root 1.72 <p>There are a number of principal ways to embed other event loops or modules
1518     into libev. Here are some ideas on how to include libadns into libev
1519     (there is a Perl module named <code>EV::ADNS</code> that does this, which you could
1520     use for an actually working example. Another Perl module named <code>EV::Glib</code>
1521     embeds a Glib main context into libev, and finally, <code>Glib::EV</code> embeds EV
1522     into the Glib event loop).</p>
1523     <p>Method 1: Add IO watchers and a timeout watcher in a prepare handler,
1524     and in a check watcher, destroy them and call into libadns. What follows
1525     is pseudo-code only of course. This requires you to either use a low
1526     priority for the check watcher or use <code>ev_clear_pending</code> explicitly, as
1527     the callbacks for the IO/timeout watchers might not have been called yet.</p>
1528 root 1.46 <pre> static ev_io iow [nfd];
1529     static ev_timer tw;
1530    
1531     static void
1532     io_cb (ev_loop *loop, ev_io *w, int revents)
1533     {
1534     }
1535    
1536     // create io watchers for each fd and a timer before blocking
1537     static void
1538     adns_prepare_cb (ev_loop *loop, ev_prepare *w, int revents)
1539     {
1540 root 1.64 int timeout = 3600000;
1541     struct pollfd fds [nfd];
1542 root 1.46 // actual code will need to loop here and realloc etc.
1543     adns_beforepoll (ads, fds, &amp;nfd, &amp;timeout, timeval_from (ev_time ()));
1544    
1545     /* the callback is illegal, but won't be called as we stop during check */
1546     ev_timer_init (&amp;tw, 0, timeout * 1e-3);
1547     ev_timer_start (loop, &amp;tw);
1548    
1549 root 1.72 // create one ev_io per pollfd
1550 root 1.46 for (int i = 0; i &lt; nfd; ++i)
1551     {
1552     ev_io_init (iow + i, io_cb, fds [i].fd,
1553     ((fds [i].events &amp; POLLIN ? EV_READ : 0)
1554     | (fds [i].events &amp; POLLOUT ? EV_WRITE : 0)));
1555    
1556     fds [i].revents = 0;
1557     ev_io_start (loop, iow + i);
1558     }
1559     }
1560    
1561     // stop all watchers after blocking
1562     static void
1563     adns_check_cb (ev_loop *loop, ev_check *w, int revents)
1564     {
1565     ev_timer_stop (loop, &amp;tw);
1566    
1567     for (int i = 0; i &lt; nfd; ++i)
1568 root 1.72 {
1569     // set the relevant poll flags
1570     // could also call adns_processreadable etc. here
1571     struct pollfd *fd = fds + i;
1572     int revents = ev_clear_pending (iow + i);
1573     if (revents &amp; EV_READ ) fd-&gt;revents |= fd-&gt;events &amp; POLLIN;
1574     if (revents &amp; EV_WRITE) fd-&gt;revents |= fd-&gt;events &amp; POLLOUT;
1575    
1576     // now stop the watcher
1577     ev_io_stop (loop, iow + i);
1578     }
1579 root 1.46
1580     adns_afterpoll (adns, fds, nfd, timeval_from (ev_now (loop));
1581     }
1582 root 1.35
1583 root 1.72 </pre>
1584     <p>Method 2: This would be just like method 1, but you run <code>adns_afterpoll</code>
1585     in the prepare watcher and would dispose of the check watcher.</p>
1586     <p>Method 3: If the module to be embedded supports explicit event
1587     notification (adns does), you can also make use of the actual watcher
1588     callbacks, and only destroy/create the watchers in the prepare watcher.</p>
1589     <pre> static void
1590     timer_cb (EV_P_ ev_timer *w, int revents)
1591     {
1592     adns_state ads = (adns_state)w-&gt;data;
1593     update_now (EV_A);
1594    
1595     adns_processtimeouts (ads, &amp;tv_now);
1596     }
1597    
1598     static void
1599     io_cb (EV_P_ ev_io *w, int revents)
1600     {
1601     adns_state ads = (adns_state)w-&gt;data;
1602     update_now (EV_A);
1603    
1604     if (revents &amp; EV_READ ) adns_processreadable (ads, w-&gt;fd, &amp;tv_now);
1605     if (revents &amp; EV_WRITE) adns_processwriteable (ads, w-&gt;fd, &amp;tv_now);
1606     }
1607    
1608     // do not ever call adns_afterpoll
1609    
1610     </pre>
1611     <p>Method 4: Do not use a prepare or check watcher because the module you
1612     want to embed is too inflexible to support it. Instead, youc na override
1613     their poll function. The drawback with this solution is that the main
1614     loop is now no longer controllable by EV. The <code>Glib::EV</code> module does
1615     this.</p>
1616     <pre> static gint
1617     event_poll_func (GPollFD *fds, guint nfds, gint timeout)
1618     {
1619     int got_events = 0;
1620    
1621     for (n = 0; n &lt; nfds; ++n)
1622     // create/start io watcher that sets the relevant bits in fds[n] and increment got_events
1623    
1624     if (timeout &gt;= 0)
1625     // create/start timer
1626    
1627     // poll
1628     ev_loop (EV_A_ 0);
1629    
1630     // stop timer again
1631     if (timeout &gt;= 0)
1632     ev_timer_stop (EV_A_ &amp;to);
1633    
1634     // stop io watchers again - their callbacks should have set
1635     for (n = 0; n &lt; nfds; ++n)
1636     ev_io_stop (EV_A_ iow [n]);
1637    
1638     return got_events;
1639     }
1640    
1641 root 1.35
1642    
1643    
1644 root 1.46 </pre>
1645 root 1.1
1646     </div>
1647 root 1.43 <h2 id="code_ev_embed_code_when_one_backend_"><code>ev_embed</code> - when one backend isn't enough...</h2>
1648 root 1.36 <div id="code_ev_embed_code_when_one_backend_-2">
1649     <p>This is a rather advanced watcher type that lets you embed one event loop
1650 root 1.37 into another (currently only <code>ev_io</code> events are supported in the embedded
1651     loop, other types of watchers might be handled in a delayed or incorrect
1652     fashion and must not be used).</p>
1653 root 1.36 <p>There are primarily two reasons you would want that: work around bugs and
1654     prioritise I/O.</p>
1655     <p>As an example for a bug workaround, the kqueue backend might only support
1656     sockets on some platform, so it is unusable as generic backend, but you
1657     still want to make use of it because you have many sockets and it scales
1658     so nicely. In this case, you would create a kqueue-based loop and embed it
1659     into your default loop (which might use e.g. poll). Overall operation will
1660     be a bit slower because first libev has to poll and then call kevent, but
1661     at least you can use both at what they are best.</p>
1662     <p>As for prioritising I/O: rarely you have the case where some fds have
1663     to be watched and handled very quickly (with low latency), and even
1664     priorities and idle watchers might have too much overhead. In this case
1665     you would put all the high priority stuff in one loop and all the rest in
1666     a second one, and embed the second one in the first.</p>
1667 root 1.37 <p>As long as the watcher is active, the callback will be invoked every time
1668     there might be events pending in the embedded loop. The callback must then
1669     call <code>ev_embed_sweep (mainloop, watcher)</code> to make a single sweep and invoke
1670     their callbacks (you could also start an idle watcher to give the embedded
1671     loop strictly lower priority for example). You can also set the callback
1672     to <code>0</code>, in which case the embed watcher will automatically execute the
1673     embedded loop sweep.</p>
1674 root 1.36 <p>As long as the watcher is started it will automatically handle events. The
1675     callback will be invoked whenever some events have been handled. You can
1676     set the callback to <code>0</code> to avoid having to specify one if you are not
1677     interested in that.</p>
1678     <p>Also, there have not currently been made special provisions for forking:
1679     when you fork, you not only have to call <code>ev_loop_fork</code> on both loops,
1680     but you will also have to stop and restart any <code>ev_embed</code> watchers
1681     yourself.</p>
1682     <p>Unfortunately, not all backends are embeddable, only the ones returned by
1683     <code>ev_embeddable_backends</code> are, which, unfortunately, does not include any
1684     portable one.</p>
1685     <p>So when you want to use this feature you will always have to be prepared
1686     that you cannot get an embeddable loop. The recommended way to get around
1687     this is to have a separate variables for your embeddable loop, try to
1688     create it, and if that fails, use the normal loop for everything:</p>
1689     <pre> struct ev_loop *loop_hi = ev_default_init (0);
1690     struct ev_loop *loop_lo = 0;
1691     struct ev_embed embed;
1692    
1693     // see if there is a chance of getting one that works
1694     // (remember that a flags value of 0 means autodetection)
1695     loop_lo = ev_embeddable_backends () &amp; ev_recommended_backends ()
1696     ? ev_loop_new (ev_embeddable_backends () &amp; ev_recommended_backends ())
1697     : 0;
1698    
1699     // if we got one, then embed it, otherwise default to loop_hi
1700     if (loop_lo)
1701     {
1702     ev_embed_init (&amp;embed, 0, loop_lo);
1703     ev_embed_start (loop_hi, &amp;embed);
1704     }
1705     else
1706     loop_lo = loop_hi;
1707    
1708     </pre>
1709     <dl>
1710 root 1.37 <dt>ev_embed_init (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1711     <dt>ev_embed_set (ev_embed *, callback, struct ev_loop *embedded_loop)</dt>
1712 root 1.36 <dd>
1713 root 1.37 <p>Configures the watcher to embed the given loop, which must be
1714     embeddable. If the callback is <code>0</code>, then <code>ev_embed_sweep</code> will be
1715     invoked automatically, otherwise it is the responsibility of the callback
1716     to invoke it (it will continue to be called until the sweep has been done,
1717     if you do not want thta, you need to temporarily stop the embed watcher).</p>
1718     </dd>
1719     <dt>ev_embed_sweep (loop, ev_embed *)</dt>
1720     <dd>
1721     <p>Make a single, non-blocking sweep over the embedded loop. This works
1722     similarly to <code>ev_loop (embedded_loop, EVLOOP_NONBLOCK)</code>, but in the most
1723     apropriate way for embedded loops.</p>
1724 root 1.36 </dd>
1725 root 1.48 <dt>struct ev_loop *loop [read-only]</dt>
1726     <dd>
1727     <p>The embedded event loop.</p>
1728     </dd>
1729 root 1.36 </dl>
1730    
1731    
1732    
1733    
1734    
1735     </div>
1736 root 1.50 <h2 id="code_ev_fork_code_the_audacity_to_re"><code>ev_fork</code> - the audacity to resume the event loop after a fork</h2>
1737     <div id="code_ev_fork_code_the_audacity_to_re-2">
1738     <p>Fork watchers are called when a <code>fork ()</code> was detected (usually because
1739     whoever is a good citizen cared to tell libev about it by calling
1740     <code>ev_default_fork</code> or <code>ev_loop_fork</code>). The invocation is done before the
1741     event loop blocks next and before <code>ev_check</code> watchers are being called,
1742     and only in the child after the fork. If whoever good citizen calling
1743     <code>ev_default_fork</code> cheats and calls it in the wrong process, the fork
1744     handlers will be invoked, too, of course.</p>
1745     <dl>
1746     <dt>ev_fork_init (ev_signal *, callback)</dt>
1747     <dd>
1748     <p>Initialises and configures the fork watcher - it has no parameters of any
1749     kind. There is a <code>ev_fork_set</code> macro, but using it is utterly pointless,
1750     believe me.</p>
1751     </dd>
1752     </dl>
1753    
1754    
1755    
1756    
1757    
1758     </div>
1759 root 1.55 <h1 id="OTHER_FUNCTIONS">OTHER FUNCTIONS</h1>
1760 root 1.1 <div id="OTHER_FUNCTIONS_CONTENT">
1761 root 1.14 <p>There are some other functions of possible interest. Described. Here. Now.</p>
1762 root 1.1 <dl>
1763     <dt>ev_once (loop, int fd, int events, ev_tstamp timeout, callback)</dt>
1764     <dd>
1765     <p>This function combines a simple timer and an I/O watcher, calls your
1766     callback on whichever event happens first and automatically stop both
1767     watchers. This is useful if you want to wait for a single event on an fd
1768 root 1.25 or timeout without having to allocate/configure/start/stop/free one or
1769 root 1.1 more watchers yourself.</p>
1770 root 1.14 <p>If <code>fd</code> is less than 0, then no I/O watcher will be started and events
1771     is being ignored. Otherwise, an <code>ev_io</code> watcher for the given <code>fd</code> and
1772     <code>events</code> set will be craeted and started.</p>
1773 root 1.1 <p>If <code>timeout</code> is less than 0, then no timeout watcher will be
1774 root 1.14 started. Otherwise an <code>ev_timer</code> watcher with after = <code>timeout</code> (and
1775     repeat = 0) will be started. While <code>0</code> is a valid timeout, it is of
1776     dubious value.</p>
1777     <p>The callback has the type <code>void (*cb)(int revents, void *arg)</code> and gets
1778 root 1.24 passed an <code>revents</code> set like normal event callbacks (a combination of
1779 root 1.14 <code>EV_ERROR</code>, <code>EV_READ</code>, <code>EV_WRITE</code> or <code>EV_TIMEOUT</code>) and the <code>arg</code>
1780     value passed to <code>ev_once</code>:</p>
1781 root 1.1 <pre> static void stdin_ready (int revents, void *arg)
1782     {
1783     if (revents &amp; EV_TIMEOUT)
1784 root 1.14 /* doh, nothing entered */;
1785 root 1.1 else if (revents &amp; EV_READ)
1786 root 1.14 /* stdin might have data for us, joy! */;
1787 root 1.1 }
1788    
1789 root 1.14 ev_once (STDIN_FILENO, EV_READ, 10., stdin_ready, 0);
1790 root 1.1
1791     </pre>
1792     </dd>
1793 root 1.37 <dt>ev_feed_event (ev_loop *, watcher *, int revents)</dt>
1794 root 1.1 <dd>
1795     <p>Feeds the given event set into the event loop, as if the specified event
1796 root 1.14 had happened for the specified watcher (which must be a pointer to an
1797     initialised but not necessarily started event watcher).</p>
1798 root 1.1 </dd>
1799 root 1.37 <dt>ev_feed_fd_event (ev_loop *, int fd, int revents)</dt>
1800 root 1.1 <dd>
1801 root 1.14 <p>Feed an event on the given fd, as if a file descriptor backend detected
1802     the given events it.</p>
1803 root 1.1 </dd>
1804 root 1.37 <dt>ev_feed_signal_event (ev_loop *loop, int signum)</dt>
1805 root 1.1 <dd>
1806 root 1.37 <p>Feed an event as if the given signal occured (<code>loop</code> must be the default
1807     loop!).</p>
1808 root 1.1 </dd>
1809     </dl>
1810    
1811 root 1.35
1812    
1813    
1814    
1815 root 1.1 </div>
1816 root 1.55 <h1 id="LIBEVENT_EMULATION">LIBEVENT EMULATION</h1>
1817 root 1.23 <div id="LIBEVENT_EMULATION_CONTENT">
1818 root 1.26 <p>Libev offers a compatibility emulation layer for libevent. It cannot
1819     emulate the internals of libevent, so here are some usage hints:</p>
1820     <dl>
1821     <dt>* Use it by including &lt;event.h&gt;, as usual.</dt>
1822     <dt>* The following members are fully supported: ev_base, ev_callback,
1823     ev_arg, ev_fd, ev_res, ev_events.</dt>
1824     <dt>* Avoid using ev_flags and the EVLIST_*-macros, while it is
1825     maintained by libev, it does not work exactly the same way as in libevent (consider
1826     it a private API).</dt>
1827     <dt>* Priorities are not currently supported. Initialising priorities
1828     will fail and all watchers will have the same priority, even though there
1829     is an ev_pri field.</dt>
1830     <dt>* Other members are not supported.</dt>
1831     <dt>* The libev emulation is <i>not</i> ABI compatible to libevent, you need
1832     to use the libev header file and library.</dt>
1833     </dl>
1834 root 1.23
1835     </div>
1836 root 1.55 <h1 id="C_SUPPORT">C++ SUPPORT</h1>
1837 root 1.23 <div id="C_SUPPORT_CONTENT">
1838 root 1.39 <p>Libev comes with some simplistic wrapper classes for C++ that mainly allow
1839     you to use some convinience methods to start/stop watchers and also change
1840     the callback model to a model using method callbacks on objects.</p>
1841     <p>To use it,</p>
1842     <pre> #include &lt;ev++.h&gt;
1843    
1844     </pre>
1845 root 1.68 <p>This automatically includes <cite>ev.h</cite> and puts all of its definitions (many
1846     of them macros) into the global namespace. All C++ specific things are
1847     put into the <code>ev</code> namespace. It should support all the same embedding
1848     options as <cite>ev.h</cite>, most notably <code>EV_MULTIPLICITY</code>.</p>
1849 root 1.69 <p>Care has been taken to keep the overhead low. The only data member the C++
1850     classes add (compared to plain C-style watchers) is the event loop pointer
1851     that the watcher is associated with (or no additional members at all if
1852     you disable <code>EV_MULTIPLICITY</code> when embedding libev).</p>
1853     <p>Currently, functions, and static and non-static member functions can be
1854 root 1.68 used as callbacks. Other types should be easy to add as long as they only
1855     need one additional pointer for context. If you need support for other
1856     types of functors please contact the author (preferably after implementing
1857     it).</p>
1858 root 1.39 <p>Here is a list of things available in the <code>ev</code> namespace:</p>
1859     <dl>
1860     <dt><code>ev::READ</code>, <code>ev::WRITE</code> etc.</dt>
1861     <dd>
1862     <p>These are just enum values with the same values as the <code>EV_READ</code> etc.
1863     macros from <cite>ev.h</cite>.</p>
1864     </dd>
1865     <dt><code>ev::tstamp</code>, <code>ev::now</code></dt>
1866     <dd>
1867     <p>Aliases to the same types/functions as with the <code>ev_</code> prefix.</p>
1868     </dd>
1869     <dt><code>ev::io</code>, <code>ev::timer</code>, <code>ev::periodic</code>, <code>ev::idle</code>, <code>ev::sig</code> etc.</dt>
1870     <dd>
1871     <p>For each <code>ev_TYPE</code> watcher in <cite>ev.h</cite> there is a corresponding class of
1872     the same name in the <code>ev</code> namespace, with the exception of <code>ev_signal</code>
1873     which is called <code>ev::sig</code> to avoid clashes with the <code>signal</code> macro
1874     defines by many implementations.</p>
1875     <p>All of those classes have these methods:</p>
1876     <p>
1877     <dl>
1878 root 1.68 <dt>ev::TYPE::TYPE ()</dt>
1879     <dt>ev::TYPE::TYPE (struct ev_loop *)</dt>
1880 root 1.39 <dt>ev::TYPE::~TYPE</dt>
1881     <dd>
1882 root 1.68 <p>The constructor (optionally) takes an event loop to associate the watcher
1883     with. If it is omitted, it will use <code>EV_DEFAULT</code>.</p>
1884     <p>The constructor calls <code>ev_init</code> for you, which means you have to call the
1885     <code>set</code> method before starting it.</p>
1886     <p>It will not set a callback, however: You have to call the templated <code>set</code>
1887     method to set a callback before you can start the watcher.</p>
1888     <p>(The reason why you have to use a method is a limitation in C++ which does
1889     not allow explicit template arguments for constructors).</p>
1890 root 1.39 <p>The destructor automatically stops the watcher if it is active.</p>
1891     </dd>
1892 root 1.68 <dt>w-&gt;set&lt;class, &amp;class::method&gt; (object *)</dt>
1893     <dd>
1894     <p>This method sets the callback method to call. The method has to have a
1895     signature of <code>void (*)(ev_TYPE &amp;, int)</code>, it receives the watcher as
1896     first argument and the <code>revents</code> as second. The object must be given as
1897     parameter and is stored in the <code>data</code> member of the watcher.</p>
1898     <p>This method synthesizes efficient thunking code to call your method from
1899     the C callback that libev requires. If your compiler can inline your
1900     callback (i.e. it is visible to it at the place of the <code>set</code> call and
1901     your compiler is good :), then the method will be fully inlined into the
1902     thunking function, making it as fast as a direct C callback.</p>
1903     <p>Example: simple class declaration and watcher initialisation</p>
1904     <pre> struct myclass
1905     {
1906     void io_cb (ev::io &amp;w, int revents) { }
1907     }
1908    
1909     myclass obj;
1910     ev::io iow;
1911     iow.set &lt;myclass, &amp;myclass::io_cb&gt; (&amp;obj);
1912    
1913     </pre>
1914     </dd>
1915 root 1.70 <dt>w-&gt;set&lt;function&gt; (void *data = 0)</dt>
1916 root 1.68 <dd>
1917     <p>Also sets a callback, but uses a static method or plain function as
1918     callback. The optional <code>data</code> argument will be stored in the watcher's
1919     <code>data</code> member and is free for you to use.</p>
1920 root 1.70 <p>The prototype of the <code>function</code> must be <code>void (*)(ev::TYPE &amp;w, int)</code>.</p>
1921 root 1.68 <p>See the method-<code>set</code> above for more details.</p>
1922 root 1.70 <p>Example:</p>
1923     <pre> static void io_cb (ev::io &amp;w, int revents) { }
1924     iow.set &lt;io_cb&gt; ();
1925    
1926     </pre>
1927 root 1.68 </dd>
1928 root 1.39 <dt>w-&gt;set (struct ev_loop *)</dt>
1929     <dd>
1930     <p>Associates a different <code>struct ev_loop</code> with this watcher. You can only
1931     do this when the watcher is inactive (and not pending either).</p>
1932     </dd>
1933     <dt>w-&gt;set ([args])</dt>
1934     <dd>
1935     <p>Basically the same as <code>ev_TYPE_set</code>, with the same args. Must be
1936 root 1.68 called at least once. Unlike the C counterpart, an active watcher gets
1937     automatically stopped and restarted when reconfiguring it with this
1938     method.</p>
1939 root 1.39 </dd>
1940     <dt>w-&gt;start ()</dt>
1941     <dd>
1942 root 1.68 <p>Starts the watcher. Note that there is no <code>loop</code> argument, as the
1943     constructor already stores the event loop.</p>
1944 root 1.39 </dd>
1945     <dt>w-&gt;stop ()</dt>
1946     <dd>
1947     <p>Stops the watcher if it is active. Again, no <code>loop</code> argument.</p>
1948     </dd>
1949     <dt>w-&gt;again () <code>ev::timer</code>, <code>ev::periodic</code> only</dt>
1950     <dd>
1951     <p>For <code>ev::timer</code> and <code>ev::periodic</code>, this invokes the corresponding
1952     <code>ev_TYPE_again</code> function.</p>
1953     </dd>
1954     <dt>w-&gt;sweep () <code>ev::embed</code> only</dt>
1955     <dd>
1956     <p>Invokes <code>ev_embed_sweep</code>.</p>
1957     </dd>
1958 root 1.49 <dt>w-&gt;update () <code>ev::stat</code> only</dt>
1959     <dd>
1960     <p>Invokes <code>ev_stat_stat</code>.</p>
1961     </dd>
1962 root 1.39 </dl>
1963     </p>
1964     </dd>
1965     </dl>
1966     <p>Example: Define a class with an IO and idle watcher, start one of them in
1967     the constructor.</p>
1968     <pre> class myclass
1969     {
1970     ev_io io; void io_cb (ev::io &amp;w, int revents);
1971     ev_idle idle void idle_cb (ev::idle &amp;w, int revents);
1972    
1973     myclass ();
1974     }
1975    
1976     myclass::myclass (int fd)
1977     {
1978 root 1.68 io .set &lt;myclass, &amp;myclass::io_cb &gt; (this);
1979     idle.set &lt;myclass, &amp;myclass::idle_cb&gt; (this);
1980    
1981 root 1.39 io.start (fd, ev::READ);
1982     }
1983    
1984 root 1.50
1985    
1986    
1987     </pre>
1988    
1989     </div>
1990 root 1.55 <h1 id="MACRO_MAGIC">MACRO MAGIC</h1>
1991 root 1.50 <div id="MACRO_MAGIC_CONTENT">
1992     <p>Libev can be compiled with a variety of options, the most fundemantal is
1993 root 1.65 <code>EV_MULTIPLICITY</code>. This option determines whether (most) functions and
1994 root 1.50 callbacks have an initial <code>struct ev_loop *</code> argument.</p>
1995     <p>To make it easier to write programs that cope with either variant, the
1996     following macros are defined:</p>
1997     <dl>
1998     <dt><code>EV_A</code>, <code>EV_A_</code></dt>
1999     <dd>
2000     <p>This provides the loop <i>argument</i> for functions, if one is required (&quot;ev
2001     loop argument&quot;). The <code>EV_A</code> form is used when this is the sole argument,
2002     <code>EV_A_</code> is used when other arguments are following. Example:</p>
2003     <pre> ev_unref (EV_A);
2004     ev_timer_add (EV_A_ watcher);
2005     ev_loop (EV_A_ 0);
2006    
2007     </pre>
2008     <p>It assumes the variable <code>loop</code> of type <code>struct ev_loop *</code> is in scope,
2009     which is often provided by the following macro.</p>
2010     </dd>
2011     <dt><code>EV_P</code>, <code>EV_P_</code></dt>
2012     <dd>
2013     <p>This provides the loop <i>parameter</i> for functions, if one is required (&quot;ev
2014     loop parameter&quot;). The <code>EV_P</code> form is used when this is the sole parameter,
2015     <code>EV_P_</code> is used when other parameters are following. Example:</p>
2016     <pre> // this is how ev_unref is being declared
2017     static void ev_unref (EV_P);
2018    
2019     // this is how you can declare your typical callback
2020     static void cb (EV_P_ ev_timer *w, int revents)
2021    
2022     </pre>
2023     <p>It declares a parameter <code>loop</code> of type <code>struct ev_loop *</code>, quite
2024     suitable for use with <code>EV_A</code>.</p>
2025     </dd>
2026     <dt><code>EV_DEFAULT</code>, <code>EV_DEFAULT_</code></dt>
2027     <dd>
2028     <p>Similar to the other two macros, this gives you the value of the default
2029     loop, if multiple loops are supported (&quot;ev loop default&quot;).</p>
2030     </dd>
2031     </dl>
2032 root 1.63 <p>Example: Declare and initialise a check watcher, utilising the above
2033 root 1.65 macros so it will work regardless of whether multiple loops are supported
2034 root 1.63 or not.</p>
2035 root 1.50 <pre> static void
2036     check_cb (EV_P_ ev_timer *w, int revents)
2037     {
2038     ev_check_stop (EV_A_ w);
2039     }
2040    
2041     ev_check check;
2042     ev_check_init (&amp;check, check_cb);
2043     ev_check_start (EV_DEFAULT_ &amp;check);
2044     ev_loop (EV_DEFAULT_ 0);
2045    
2046 root 1.39 </pre>
2047 root 1.23
2048     </div>
2049 root 1.55 <h1 id="EMBEDDING">EMBEDDING</h1>
2050 root 1.40 <div id="EMBEDDING_CONTENT">
2051     <p>Libev can (and often is) directly embedded into host
2052     applications. Examples of applications that embed it include the Deliantra
2053     Game Server, the EV perl module, the GNU Virtual Private Ethernet (gvpe)
2054     and rxvt-unicode.</p>
2055     <p>The goal is to enable you to just copy the neecssary files into your
2056     source directory without having to change even a single line in them, so
2057     you can easily upgrade by simply copying (or having a checked-out copy of
2058     libev somewhere in your source tree).</p>
2059    
2060     </div>
2061     <h2 id="FILESETS">FILESETS</h2>
2062     <div id="FILESETS_CONTENT">
2063     <p>Depending on what features you need you need to include one or more sets of files
2064     in your app.</p>
2065    
2066     </div>
2067     <h3 id="CORE_EVENT_LOOP">CORE EVENT LOOP</h3>
2068     <div id="CORE_EVENT_LOOP_CONTENT">
2069     <p>To include only the libev core (all the <code>ev_*</code> functions), with manual
2070     configuration (no autoconf):</p>
2071     <pre> #define EV_STANDALONE 1
2072     #include &quot;ev.c&quot;
2073    
2074     </pre>
2075     <p>This will automatically include <cite>ev.h</cite>, too, and should be done in a
2076     single C source file only to provide the function implementations. To use
2077     it, do the same for <cite>ev.h</cite> in all files wishing to use this API (best
2078     done by writing a wrapper around <cite>ev.h</cite> that you can include instead and
2079     where you can put other configuration options):</p>
2080     <pre> #define EV_STANDALONE 1
2081     #include &quot;ev.h&quot;
2082    
2083     </pre>
2084     <p>Both header files and implementation files can be compiled with a C++
2085     compiler (at least, thats a stated goal, and breakage will be treated
2086     as a bug).</p>
2087     <p>You need the following files in your source tree, or in a directory
2088     in your include path (e.g. in libev/ when using -Ilibev):</p>
2089     <pre> ev.h
2090     ev.c
2091     ev_vars.h
2092     ev_wrap.h
2093    
2094     ev_win32.c required on win32 platforms only
2095    
2096 root 1.63 ev_select.c only when select backend is enabled (which is enabled by default)
2097 root 1.40 ev_poll.c only when poll backend is enabled (disabled by default)
2098     ev_epoll.c only when the epoll backend is enabled (disabled by default)
2099     ev_kqueue.c only when the kqueue backend is enabled (disabled by default)
2100     ev_port.c only when the solaris port backend is enabled (disabled by default)
2101    
2102     </pre>
2103     <p><cite>ev.c</cite> includes the backend files directly when enabled, so you only need
2104 root 1.44 to compile this single file.</p>
2105 root 1.40
2106     </div>
2107     <h3 id="LIBEVENT_COMPATIBILITY_API">LIBEVENT COMPATIBILITY API</h3>
2108     <div id="LIBEVENT_COMPATIBILITY_API_CONTENT">
2109     <p>To include the libevent compatibility API, also include:</p>
2110     <pre> #include &quot;event.c&quot;
2111    
2112     </pre>
2113     <p>in the file including <cite>ev.c</cite>, and:</p>
2114     <pre> #include &quot;event.h&quot;
2115    
2116     </pre>
2117     <p>in the files that want to use the libevent API. This also includes <cite>ev.h</cite>.</p>
2118     <p>You need the following additional files for this:</p>
2119     <pre> event.h
2120     event.c
2121    
2122     </pre>
2123    
2124     </div>
2125     <h3 id="AUTOCONF_SUPPORT">AUTOCONF SUPPORT</h3>
2126     <div id="AUTOCONF_SUPPORT_CONTENT">
2127     <p>Instead of using <code>EV_STANDALONE=1</code> and providing your config in
2128     whatever way you want, you can also <code>m4_include([libev.m4])</code> in your
2129 root 1.44 <cite>configure.ac</cite> and leave <code>EV_STANDALONE</code> undefined. <cite>ev.c</cite> will then
2130     include <cite>config.h</cite> and configure itself accordingly.</p>
2131 root 1.40 <p>For this of course you need the m4 file:</p>
2132     <pre> libev.m4
2133    
2134     </pre>
2135    
2136     </div>
2137     <h2 id="PREPROCESSOR_SYMBOLS_MACROS">PREPROCESSOR SYMBOLS/MACROS</h2>
2138     <div id="PREPROCESSOR_SYMBOLS_MACROS_CONTENT">
2139     <p>Libev can be configured via a variety of preprocessor symbols you have to define
2140     before including any of its files. The default is not to build for multiplicity
2141     and only include the select backend.</p>
2142     <dl>
2143     <dt>EV_STANDALONE</dt>
2144     <dd>
2145     <p>Must always be <code>1</code> if you do not use autoconf configuration, which
2146     keeps libev from including <cite>config.h</cite>, and it also defines dummy
2147     implementations for some libevent functions (such as logging, which is not
2148     supported). It will also not define any of the structs usually found in
2149     <cite>event.h</cite> that are not directly supported by the libev core alone.</p>
2150     </dd>
2151     <dt>EV_USE_MONOTONIC</dt>
2152     <dd>
2153     <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2154     monotonic clock option at both compiletime and runtime. Otherwise no use
2155     of the monotonic clock option will be attempted. If you enable this, you
2156     usually have to link against librt or something similar. Enabling it when
2157     the functionality isn't available is safe, though, althoguh you have
2158     to make sure you link against any libraries where the <code>clock_gettime</code>
2159     function is hiding in (often <cite>-lrt</cite>).</p>
2160     </dd>
2161     <dt>EV_USE_REALTIME</dt>
2162     <dd>
2163     <p>If defined to be <code>1</code>, libev will try to detect the availability of the
2164     realtime clock option at compiletime (and assume its availability at
2165     runtime if successful). Otherwise no use of the realtime clock option will
2166     be attempted. This effectively replaces <code>gettimeofday</code> by <code>clock_get
2167     (CLOCK_REALTIME, ...)</code> and will not normally affect correctness. See tzhe note about libraries
2168     in the description of <code>EV_USE_MONOTONIC</code>, though.</p>
2169     </dd>
2170     <dt>EV_USE_SELECT</dt>
2171     <dd>
2172     <p>If undefined or defined to be <code>1</code>, libev will compile in support for the
2173     <code>select</code>(2) backend. No attempt at autodetection will be done: if no
2174     other method takes over, select will be it. Otherwise the select backend
2175     will not be compiled in.</p>
2176     </dd>
2177     <dt>EV_SELECT_USE_FD_SET</dt>
2178     <dd>
2179     <p>If defined to <code>1</code>, then the select backend will use the system <code>fd_set</code>
2180     structure. This is useful if libev doesn't compile due to a missing
2181     <code>NFDBITS</code> or <code>fd_mask</code> definition or it misguesses the bitset layout on
2182     exotic systems. This usually limits the range of file descriptors to some
2183     low limit such as 1024 or might have other limitations (winsocket only
2184     allows 64 sockets). The <code>FD_SETSIZE</code> macro, set before compilation, might
2185     influence the size of the <code>fd_set</code> used.</p>
2186     </dd>
2187     <dt>EV_SELECT_IS_WINSOCKET</dt>
2188     <dd>
2189     <p>When defined to <code>1</code>, the select backend will assume that
2190     select/socket/connect etc. don't understand file descriptors but
2191     wants osf handles on win32 (this is the case when the select to
2192     be used is the winsock select). This means that it will call
2193     <code>_get_osfhandle</code> on the fd to convert it to an OS handle. Otherwise,
2194     it is assumed that all these functions actually work on fds, even
2195     on win32. Should not be defined on non-win32 platforms.</p>
2196     </dd>
2197     <dt>EV_USE_POLL</dt>
2198     <dd>
2199     <p>If defined to be <code>1</code>, libev will compile in support for the <code>poll</code>(2)
2200     backend. Otherwise it will be enabled on non-win32 platforms. It
2201     takes precedence over select.</p>
2202     </dd>
2203     <dt>EV_USE_EPOLL</dt>
2204     <dd>
2205     <p>If defined to be <code>1</code>, libev will compile in support for the Linux
2206     <code>epoll</code>(7) backend. Its availability will be detected at runtime,
2207     otherwise another method will be used as fallback. This is the
2208     preferred backend for GNU/Linux systems.</p>
2209     </dd>
2210     <dt>EV_USE_KQUEUE</dt>
2211     <dd>
2212     <p>If defined to be <code>1</code>, libev will compile in support for the BSD style
2213     <code>kqueue</code>(2) backend. Its actual availability will be detected at runtime,
2214     otherwise another method will be used as fallback. This is the preferred
2215     backend for BSD and BSD-like systems, although on most BSDs kqueue only
2216     supports some types of fds correctly (the only platform we found that
2217     supports ptys for example was NetBSD), so kqueue might be compiled in, but
2218     not be used unless explicitly requested. The best way to use it is to find
2219 root 1.42 out whether kqueue supports your type of fd properly and use an embedded
2220 root 1.40 kqueue loop.</p>
2221     </dd>
2222     <dt>EV_USE_PORT</dt>
2223     <dd>
2224     <p>If defined to be <code>1</code>, libev will compile in support for the Solaris
2225     10 port style backend. Its availability will be detected at runtime,
2226     otherwise another method will be used as fallback. This is the preferred
2227     backend for Solaris 10 systems.</p>
2228     </dd>
2229     <dt>EV_USE_DEVPOLL</dt>
2230     <dd>
2231     <p>reserved for future expansion, works like the USE symbols above.</p>
2232     </dd>
2233 root 1.57 <dt>EV_USE_INOTIFY</dt>
2234     <dd>
2235     <p>If defined to be <code>1</code>, libev will compile in support for the Linux inotify
2236     interface to speed up <code>ev_stat</code> watchers. Its actual availability will
2237     be detected at runtime.</p>
2238     </dd>
2239 root 1.40 <dt>EV_H</dt>
2240     <dd>
2241     <p>The name of the <cite>ev.h</cite> header file used to include it. The default if
2242     undefined is <code>&lt;ev.h&gt;</code> in <cite>event.h</cite> and <code>&quot;ev.h&quot;</code> in <cite>ev.c</cite>. This
2243     can be used to virtually rename the <cite>ev.h</cite> header file in case of conflicts.</p>
2244     </dd>
2245     <dt>EV_CONFIG_H</dt>
2246     <dd>
2247     <p>If <code>EV_STANDALONE</code> isn't <code>1</code>, this variable can be used to override
2248     <cite>ev.c</cite>'s idea of where to find the <cite>config.h</cite> file, similarly to
2249     <code>EV_H</code>, above.</p>
2250     </dd>
2251     <dt>EV_EVENT_H</dt>
2252     <dd>
2253     <p>Similarly to <code>EV_H</code>, this macro can be used to override <cite>event.c</cite>'s idea
2254     of how the <cite>event.h</cite> header can be found.</p>
2255     </dd>
2256     <dt>EV_PROTOTYPES</dt>
2257     <dd>
2258     <p>If defined to be <code>0</code>, then <cite>ev.h</cite> will not define any function
2259     prototypes, but still define all the structs and other symbols. This is
2260     occasionally useful if you want to provide your own wrapper functions
2261     around libev functions.</p>
2262     </dd>
2263     <dt>EV_MULTIPLICITY</dt>
2264     <dd>
2265     <p>If undefined or defined to <code>1</code>, then all event-loop-specific functions
2266     will have the <code>struct ev_loop *</code> as first argument, and you can create
2267     additional independent event loops. Otherwise there will be no support
2268     for multiple event loops and there is no first event loop pointer
2269     argument. Instead, all functions act on the single default loop.</p>
2270     </dd>
2271 root 1.66 <dt>EV_MINPRI</dt>
2272     <dt>EV_MAXPRI</dt>
2273     <dd>
2274     <p>The range of allowed priorities. <code>EV_MINPRI</code> must be smaller or equal to
2275     <code>EV_MAXPRI</code>, but otherwise there are no non-obvious limitations. You can
2276     provide for more priorities by overriding those symbols (usually defined
2277     to be <code>-2</code> and <code>2</code>, respectively).</p>
2278     <p>When doing priority-based operations, libev usually has to linearly search
2279     all the priorities, so having many of them (hundreds) uses a lot of space
2280     and time, so using the defaults of five priorities (-2 .. +2) is usually
2281     fine.</p>
2282     <p>If your embedding app does not need any priorities, defining these both to
2283     <code>0</code> will save some memory and cpu.</p>
2284     </dd>
2285 root 1.48 <dt>EV_PERIODIC_ENABLE</dt>
2286     <dd>
2287     <p>If undefined or defined to be <code>1</code>, then periodic timers are supported. If
2288     defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2289     code.</p>
2290     </dd>
2291 root 1.64 <dt>EV_IDLE_ENABLE</dt>
2292     <dd>
2293     <p>If undefined or defined to be <code>1</code>, then idle watchers are supported. If
2294     defined to be <code>0</code>, then they are not. Disabling them saves a few kB of
2295     code.</p>
2296     </dd>
2297 root 1.48 <dt>EV_EMBED_ENABLE</dt>
2298     <dd>
2299     <p>If undefined or defined to be <code>1</code>, then embed watchers are supported. If
2300     defined to be <code>0</code>, then they are not.</p>
2301     </dd>
2302     <dt>EV_STAT_ENABLE</dt>
2303     <dd>
2304     <p>If undefined or defined to be <code>1</code>, then stat watchers are supported. If
2305     defined to be <code>0</code>, then they are not.</p>
2306     </dd>
2307 root 1.50 <dt>EV_FORK_ENABLE</dt>
2308     <dd>
2309     <p>If undefined or defined to be <code>1</code>, then fork watchers are supported. If
2310     defined to be <code>0</code>, then they are not.</p>
2311     </dd>
2312 root 1.48 <dt>EV_MINIMAL</dt>
2313 root 1.40 <dd>
2314 root 1.48 <p>If you need to shave off some kilobytes of code at the expense of some
2315     speed, define this symbol to <code>1</code>. Currently only used for gcc to override
2316     some inlining decisions, saves roughly 30% codesize of amd64.</p>
2317 root 1.40 </dd>
2318 root 1.51 <dt>EV_PID_HASHSIZE</dt>
2319     <dd>
2320     <p><code>ev_child</code> watchers use a small hash table to distribute workload by
2321     pid. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>), usually more
2322     than enough. If you need to manage thousands of children you might want to
2323 root 1.57 increase this value (<i>must</i> be a power of two).</p>
2324     </dd>
2325     <dt>EV_INOTIFY_HASHSIZE</dt>
2326     <dd>
2327     <p><code>ev_staz</code> watchers use a small hash table to distribute workload by
2328     inotify watch id. The default size is <code>16</code> (or <code>1</code> with <code>EV_MINIMAL</code>),
2329     usually more than enough. If you need to manage thousands of <code>ev_stat</code>
2330     watchers you might want to increase this value (<i>must</i> be a power of
2331     two).</p>
2332 root 1.51 </dd>
2333 root 1.40 <dt>EV_COMMON</dt>
2334     <dd>
2335     <p>By default, all watchers have a <code>void *data</code> member. By redefining
2336     this macro to a something else you can include more and other types of
2337     members. You have to define it each time you include one of the files,
2338     though, and it must be identical each time.</p>
2339     <p>For example, the perl EV module uses something like this:</p>
2340     <pre> #define EV_COMMON \
2341     SV *self; /* contains this struct */ \
2342     SV *cb_sv, *fh /* note no trailing &quot;;&quot; */
2343    
2344     </pre>
2345     </dd>
2346 root 1.45 <dt>EV_CB_DECLARE (type)</dt>
2347     <dt>EV_CB_INVOKE (watcher, revents)</dt>
2348     <dt>ev_set_cb (ev, cb)</dt>
2349 root 1.40 <dd>
2350     <p>Can be used to change the callback member declaration in each watcher,
2351     and the way callbacks are invoked and set. Must expand to a struct member
2352     definition and a statement, respectively. See the <cite>ev.v</cite> header file for
2353     their default definitions. One possible use for overriding these is to
2354 root 1.45 avoid the <code>struct ev_loop *</code> as first argument in all cases, or to use
2355     method calls instead of plain function calls in C++.</p>
2356 root 1.40
2357     </div>
2358     <h2 id="EXAMPLES">EXAMPLES</h2>
2359     <div id="EXAMPLES_CONTENT">
2360     <p>For a real-world example of a program the includes libev
2361     verbatim, you can have a look at the EV perl module
2362     (<a href="http://software.schmorp.de/pkg/EV.html">http://software.schmorp.de/pkg/EV.html</a>). It has the libev files in
2363     the <cite>libev/</cite> subdirectory and includes them in the <cite>EV/EVAPI.h</cite> (public
2364     interface) and <cite>EV.xs</cite> (implementation) files. Only the <cite>EV.xs</cite> file
2365     will be compiled. It is pretty complex because it provides its own header
2366     file.</p>
2367     <p>The usage in rxvt-unicode is simpler. It has a <cite>ev_cpp.h</cite> header file
2368 root 1.63 that everybody includes and which overrides some configure choices:</p>
2369     <pre> #define EV_MINIMAL 1
2370     #define EV_USE_POLL 0
2371 root 1.41 #define EV_MULTIPLICITY 0
2372 root 1.63 #define EV_PERIODIC_ENABLE 0
2373     #define EV_STAT_ENABLE 0
2374     #define EV_FORK_ENABLE 0
2375 root 1.41 #define EV_CONFIG_H &lt;config.h&gt;
2376 root 1.63 #define EV_MINPRI 0
2377     #define EV_MAXPRI 0
2378 root 1.40
2379 root 1.41 #include &quot;ev++.h&quot;
2380 root 1.40
2381     </pre>
2382     <p>And a <cite>ev_cpp.C</cite> implementation file that contains libev proper and is compiled:</p>
2383 root 1.41 <pre> #include &quot;ev_cpp.h&quot;
2384     #include &quot;ev.c&quot;
2385 root 1.40
2386 root 1.47
2387    
2388    
2389 root 1.40 </pre>
2390    
2391     </div>
2392 root 1.55 <h1 id="COMPLEXITIES">COMPLEXITIES</h1>
2393 root 1.47 <div id="COMPLEXITIES_CONTENT">
2394     <p>In this section the complexities of (many of) the algorithms used inside
2395     libev will be explained. For complexity discussions about backends see the
2396     documentation for <code>ev_default_init</code>.</p>
2397 root 1.67 <p>All of the following are about amortised time: If an array needs to be
2398     extended, libev needs to realloc and move the whole array, but this
2399     happens asymptotically never with higher number of elements, so O(1) might
2400     mean it might do a lengthy realloc operation in rare cases, but on average
2401     it is much faster and asymptotically approaches constant time.</p>
2402 root 1.47 <p>
2403     <dl>
2404     <dt>Starting and stopping timer/periodic watchers: O(log skipped_other_timers)</dt>
2405 root 1.66 <dd>
2406     <p>This means that, when you have a watcher that triggers in one hour and
2407     there are 100 watchers that would trigger before that then inserting will
2408     have to skip those 100 watchers.</p>
2409     </dd>
2410 root 1.47 <dt>Changing timer/periodic watchers (by autorepeat, again): O(log skipped_other_timers)</dt>
2411 root 1.66 <dd>
2412     <p>That means that for changing a timer costs less than removing/adding them
2413     as only the relative motion in the event queue has to be paid for.</p>
2414     </dd>
2415 root 1.47 <dt>Starting io/check/prepare/idle/signal/child watchers: O(1)</dt>
2416 root 1.66 <dd>
2417 root 1.67 <p>These just add the watcher into an array or at the head of a list.
2418     =item Stopping check/prepare/idle watchers: O(1)</p>
2419 root 1.66 </dd>
2420 root 1.57 <dt>Stopping an io/signal/child watcher: O(number_of_watchers_for_this_(fd/signal/pid % EV_PID_HASHSIZE))</dt>
2421 root 1.66 <dd>
2422     <p>These watchers are stored in lists then need to be walked to find the
2423     correct watcher to remove. The lists are usually short (you don't usually
2424     have many watchers waiting for the same fd or signal).</p>
2425     </dd>
2426 root 1.47 <dt>Finding the next timer per loop iteration: O(1)</dt>
2427     <dt>Each change on a file descriptor per loop iteration: O(number_of_watchers_for_this_fd)</dt>
2428 root 1.66 <dd>
2429     <p>A change means an I/O watcher gets started or stopped, which requires
2430     libev to recalculate its status (and possibly tell the kernel).</p>
2431     </dd>
2432 root 1.47 <dt>Activating one watcher: O(1)</dt>
2433 root 1.66 <dt>Priority handling: O(number_of_priorities)</dt>
2434     <dd>
2435     <p>Priorities are implemented by allocating some space for each
2436     priority. When doing priority-based operations, libev usually has to
2437     linearly search all the priorities.</p>
2438     </dd>
2439 root 1.47 </dl>
2440     </p>
2441    
2442    
2443    
2444    
2445    
2446     </div>
2447 root 1.55 <h1 id="AUTHOR">AUTHOR</h1>
2448 root 1.1 <div id="AUTHOR_CONTENT">
2449 root 1.40 <p>Marc Lehmann &lt;libev@schmorp.de&gt;.</p>
2450 root 1.1
2451     </div>
2452     </div></body>
2453     </html>